From proski at gnu.org Wed Oct 2 02:55:01 2002 From: proski at gnu.org (Pavel Roskin) Date: Tue, 1 Oct 2002 22:55:01 -0400 (EDT) Subject: problem with nls on AIX In-Reply-To: Message-ID: Hello! > > > I suspect the problem could be with gettext.0.11.5.tar.Z package from > > > aixpdslib.seas.ucla.edu (earlier I was using gettext.0.11.1 and > > > it was OK), but looking at /usr/local/lib/libintl.a or > > > mc-build.nls/intl/libintl.a I see these symbols (but with one dot > > > before, not 2, i.e. ".libintl_gettext") inside. > > > > Could you please try "--with-included-gettext" option? > > I used this option with autogen.sh, is it OK? As we can see below, it is still not OK. Please remove the gettext package completely, including the headers, and try the latest mc snapshot. > I get the same error, even after removing /usr/local/lib/libintl.a. Possible causes: 1) Your "make" is buggy and didn't recompile everything. 2) There is a problem in the headers of the gettext package. 3) There is some other library (i.e. iconv) that causes the trouble. 4) There is a bug in GNU gettext, which causes it not to disable redirection even if "--with-included-gettext" is specified. > Just FYI I here is the linking line: > > gcc -I/usr/local/include/glib-1.2 -I/usr/local/lib/glib/include > -g -O2 -Wall -o mc achown.o background.o boxes.o chmod.o chown.o cmd.o > color.o command.o complete.o cons.handler.o dialog.o dir.o dlg.o ext.o > file.o filegui.o filenot.o fileopctx.o find.o findme.o fsusage.o help.o > hotlist.o info.o key.o keyxdef.o layout.o learn.o listmode.o main.o > menu.o mountlist.o mouse.o option.o panelize.o popt.o poptconfig.o > popthelp.o poptparse.o profile.o regex.o rxvt.o screen.o setup.o slint.o > subshell.o terms.o text.o textconf.o tree.o treestore.o user.o util.o > utilunix.o view.o widget.o win.o wtools.o ../edit/libedit.a > ../slang/libmcslang.a ../vfs/libvfs-mc.a ../intl/libintl.a -liconv > -L/usr/local/lib -lglib -lcurses -liconv Where does libiconv come from? > collect2: ld returned 8 exit status > ld: 0711-317 ERROR: Undefined symbol: ..libintl_gettext > ld: 0711-317 ERROR: Undefined symbol: ..libintl_bindtextdomain > ld: 0711-317 ERROR: Undefined symbol: ..libintl_textdomain > ld: 0711-345 Use the -bloadmap or -bnoquiet option to obtain more information. What do those options show? My guess is that the unresolved symbols come from intl/intl-compat.c It would be interesting to see the output of configure and the output of "nm intl/intl-compat.o" Actually, this discussion belongs to the gettext mailing list. I don't see anything mc-specific here. If you do, then please explain. -- Regards, Pavel Roskin From ms181097 at students.mimuw.edu.pl Thu Oct 3 14:01:00 2002 From: ms181097 at students.mimuw.edu.pl (Marcin Stefaniak) Date: Thu, 3 Oct 2002 16:01:00 +0200 (CEST) Subject: auto-fill feature in mcedit Message-ID: Hi, I would like to add a feature to mcedit - internal Midnight Commander editor. The feature is: when user press Ctrl+Space, it tries to finish the word the user has just partially written, finding suitable word(s) from the edited text itself. I know this feature from JCreator Lite - freeware Java IDE for Windows - and I find this very helpful when programming (think of repeating identifiers...) no matter what proramming language is used. I heard that such a feature is also in Emacs, but my favorite editor is mcedit. Please tell me if this is already working in mcedit, or if you consider it a drawback, or hard to implement for some reasons unknown to me. I reckon I can modify mcedit, it won't be difficult, will it? best regards, Marcin Stefaniak matmis at mimuw.edu.pl From kai at cmail.ru Thu Oct 3 15:19:04 2002 From: kai at cmail.ru (Andrew V. Samoilov) Date: Thu, 03 Oct 2002 18:19:04 +0300 Subject: auto-fill feature in mcedit References: Message-ID: <3D9C5FE8.1070906@cmail.ru> Marcin Stefaniak wrote: > Hi, > > I would like to add a feature to mcedit - internal Midnight Commander > editor. The feature is: when user press Ctrl+Space, it tries to finish the > word the user has just partially written, finding suitable word(s) from > the edited text itself. This feature is already implemented in the current mc. You need to press Escape Tab. Current code snapshot and rpm for Redhat 7.x (i386) can be downloaded from http://www.ibiblio.org/pub/Linux/utils/file/managers/mc/snapshots/ -- Regards, Andrew V. Samoilov From proski at gnu.org Thu Oct 3 15:44:10 2002 From: proski at gnu.org (Pavel Roskin) Date: Thu, 3 Oct 2002 11:44:10 -0400 (EDT) Subject: auto-fill feature in mcedit In-Reply-To: <3D9C5FE8.1070906@cmail.ru> Message-ID: > Current code snapshot and rpm for Redhat 7.x (i386) can be downloaded from > http://www.ibiblio.org/pub/Linux/utils/file/managers/mc/snapshots/ Sorry, it's RedHat 8.0 packages since yesterday. However, users of other RPM-based distributions can use "rpm -tb mc*.tar.gz" or "rpmbuild -tb mc*.tar.gz" to build the binary package. The S-Lang library for RedHat 8.0 is compiled with UTF support and causes some minor problems on the Linux console. I'm considering using the internal S-Lang in the RPM packages again (they are just for testing, so less dependencies is a good thing), but only if we fix cut-and-paste (internal S-Lang fills the whole screen with spaces). Escape-Tab (or Alt-Tab if it's not consumed by the window manager) in the editor is present in the 4.6.0-pre1 pre-release as well. -- Regards, Pavel Roskin From kai at cmail.ru Thu Oct 3 17:17:01 2002 From: kai at cmail.ru (Andrew V. Samoilov) Date: Thu, 03 Oct 2002 20:17:01 +0300 Subject: popt and i18n Message-ID: <3D9C7B8D.1000602@cmail.ru> Hello! There is POPT_ macro used in the popt subsystem for translated strings. This macro is not recognized by xgettext, so I see 2 ways to see these strings accepted: 1) Replace POPT_ with _ in src/popt*.* 2) Add POPT_ to the po/Makevars. This way requires gettext 0.11+ to be used. -- Regards, Andrew V. Samoilov From kai at cmail.ru Thu Oct 3 17:33:09 2002 From: kai at cmail.ru (Andrew V. Samoilov) Date: Thu, 03 Oct 2002 20:33:09 +0300 Subject: mc_tmpdir(), system recovery and read-only /tmp Message-ID: <3D9C7F55.8070209@cmail.ru> Hello! Recent patch makes mc unusable for system recovery when / is mounted read-only, and there is no /tmp/mc-root. Also it does not check ownership and modes if such directory existed, and so it can be used for some kinds of DOS in mc. -- Regards, Andrew V. Samoilov From proski at gnu.org Thu Oct 3 17:40:22 2002 From: proski at gnu.org (Pavel Roskin) Date: Thu, 3 Oct 2002 13:40:22 -0400 (EDT) Subject: popt and i18n In-Reply-To: <3D9C7B8D.1000602@cmail.ru> Message-ID: On Thu, 3 Oct 2002, Andrew V. Samoilov wrote: > Hello! > > There is POPT_ macro used in the popt subsystem for translated strings. > This macro is not recognized by xgettext, so I see 2 ways to see these > strings accepted: > > 1) Replace POPT_ with _ in src/popt*.* > 2) Add POPT_ to the po/Makevars. This way requires gettext 0.11+ to be > used. Can you actually make one of those messages appear on the screen? I guess the reason why POPT_ is used is to save translators time translating things that have very little chance to appear on the screen, at least for "normal" users. -- Regards, Pavel Roskin From kai at cmail.ru Thu Oct 3 17:50:17 2002 From: kai at cmail.ru (Andrew V. Samoilov) Date: Thu, 03 Oct 2002 20:50:17 +0300 Subject: popt and i18n References: Message-ID: <3D9C8359.1080100@cmail.ru> Pavel Roskin wrote: > On Thu, 3 Oct 2002, Andrew V. Samoilov wrote: > > >>Hello! >> >>There is POPT_ macro used in the popt subsystem for translated strings. >> This macro is not recognized by xgettext, so I see 2 ways to see these >>strings accepted: >> >>1) Replace POPT_ with _ in src/popt*.* >>2) Add POPT_ to the po/Makevars. This way requires gettext 0.11+ to be >>used. > > > Can you actually make one of those messages appear on the screen? I guess > the reason why POPT_ is used is to save translators time translating > things that have very little chance to appear on the screen, at least for > "normal" users. > mc -? and you can see at least 4 such messages: "Usage:", "[OPTION...]", "ARG" ;-) and "-?: unknown option". -- Regards, Andrew V. Samoilov From proski at gnu.org Thu Oct 3 19:00:12 2002 From: proski at gnu.org (Pavel Roskin) Date: Thu, 3 Oct 2002 15:00:12 -0400 (EDT) Subject: popt and i18n In-Reply-To: <3D9C8359.1080100@cmail.ru> Message-ID: > mc -? > and you can see at least 4 such messages: "Usage:", "[OPTION...]", "ARG" > ;-) and "-?: unknown option". Well, I cannot. It seems that libc locale support is broken in Red Hat 8.0, so all the messages are in English. I'll try to fix all this mess as soon as possible :-( I think that "POPT_" should be replaced with "_". There are already many local changes to popt. If we decide to update popt at a later date, then we can think if it's OK to obsolete gettext 0.10.x. -- Regards, Pavel Roskin From proski at gnu.org Thu Oct 3 19:01:18 2002 From: proski at gnu.org (Pavel Roskin) Date: Thu, 3 Oct 2002 15:01:18 -0400 (EDT) Subject: mc_tmpdir(), system recovery and read-only /tmp In-Reply-To: <3D9C7F55.8070209@cmail.ru> Message-ID: > Recent patch makes mc unusable for system recovery when / is mounted > read-only, and there is no /tmp/mc-root. Also it does not check > ownership and modes if such directory existed, and so it can be used for > some kinds of DOS in mc. Indeed, that should be fixed. I have added this to src/TODO and marked it as a regression. -- Regards, Pavel Roskin From spambox at centrum.cz Fri Oct 4 13:51:03 2002 From: spambox at centrum.cz (Jiri Mencak) Date: Fri, 4 Oct 2002 14:51:03 +0100 Subject: copy_current_pathname() Message-ID: <20021004135103.GA30648@mencak.homeip.net> Hello, when copying current pathname containing characters like [\'\\\n\t\r":; ] (and many others---for a full list see `util.c', name_quote()) on the MC's command line, the function name_quote() is not used. Is this intentional? If not, the patch for MC 4.6.0-pre1 is attached. I've basically lifted it from copy_prog_name() in `main.c'. On another note, for those who remember the recent thread "Keyboard configurability", unfortunately, multiple escape sequences for the same keys in `~/.mc/ini' in section [terminal:XXX] didn't work for me. I had to hardcode them in the source. This is, of course, a temporary workaround. When I have more time, I'll try to find out why it didn't work. Regards, Jiri PS: Trying to be responsible ;) -------------- next part -------------- --- main.c.orig Wed Aug 21 02:08:49 2002 +++ main.c Fri Oct 4 13:21:30 2002 @@ -1258,12 +1258,16 @@ static void copy_current_pathname (void) { + char *tmp; if (!command_prompt) return; - stuff (input_w (cmdline), cpanel->cwd, 0); - if (cpanel->cwd [strlen (cpanel->cwd) - 1] != PATH_SEP) + tmp = name_quote (cpanel->cwd, 1); + stuff (input_w (cmdline), tmp, 0); + g_free (tmp); + if (cpanel->cwd [strlen (cpanel->cwd) - 1] != PATH_SEP) { stuff (input_w (cmdline), PATH_SEP_STR, 0); + } } static void copy_other_pathname (void) From proski at gnu.org Fri Oct 4 20:46:19 2002 From: proski at gnu.org (Pavel Roskin) Date: Fri, 4 Oct 2002 16:46:19 -0400 (EDT) Subject: copy_current_pathname() In-Reply-To: <20021004135103.GA30648@mencak.homeip.net> Message-ID: Hello, Jiri! > when copying current pathname containing characters like > [\'\\\n\t\r":; ] (and many others---for a full list see `util.c', > name_quote()) on the MC's command line, the function name_quote() > is not used. Is this intentional? If not, the patch for > MC 4.6.0-pre1 is attached. I believe it's fixed already in the current snapshot. -- Regards, Pavel Roskin From pehrens at ligo.caltech.edu Fri Oct 4 21:09:45 2002 From: pehrens at ligo.caltech.edu (Philip Ehrens) Date: Fri, 4 Oct 2002 14:09:45 -0700 Subject: executable scripts in /tmp Message-ID: <20021004210945.GC5614@ligo.caltech.edu> Hi Folks, mc currently writes executable scripts into /tmp. I think this is something that should generally be avoided. On my pokey little server I have /tmp mounted noexec because I am paranoid. Why, since ~/.mc/tmp exists, and is used by mc anyway, doesn't mc just use ~/.mc/tmp for everything? I have patched it to do this (by editing mc.ext and starting mc with "/usr/bin/env TMPDIR=~/.mc/tmp") and it works splendidly this way. Of course the proper fix simply requires replacing TMPDIR with HOME/.mc/tmp in the right place(s). Is there a reason it does not do this by default? Should I submit a proper patch? Inquiring minds want to gnow. Phil -- Phil Ehrens | Fun stuff: The LIGO Laboratory, MS 18-34 | http://www.ralphmag.org California Institute of Technology | http://www.yellow5.com 1200 East California Blvd. | http://www.total.net/~fishnet/ Pasadena, CA 91125 USA | http://slashdot.org Phone:(626)395-8518 Fax:(626)793-9744 | http://kame56.homepage.com From proski at gnu.org Fri Oct 4 22:46:09 2002 From: proski at gnu.org (Pavel Roskin) Date: Fri, 4 Oct 2002 18:46:09 -0400 (EDT) Subject: executable scripts in /tmp In-Reply-To: <20021004210945.GC5614@ligo.caltech.edu> Message-ID: Hi, Philip! > mc currently writes executable scripts into /tmp. I'll fix it. Added to src/TODO and committed. > I think this is something that should generally be avoided. I haven't heard that it should be avoided in general (apart from portability to 20 years old UNIX without "#!" support). Any references? > On my pokey little server I have /tmp mounted noexec because > I am paranoid. OK, reliance on the ability to execute scripts seems unnecessary. > Why, since ~/.mc/tmp exists, and is used by mc anyway, doesn't > mc just use ~/.mc/tmp for everything? There is no reason whatsoever to believe that the home directory is less likely to be mounted noexec than /tmp. Really paranoid sysadmins would make both noexec :-) On the other hand, I've seen installations where the home directory is mounted over NFS and shared between several machines. That's not a good choice for temporary files. ~/.mc/tmp is only used by the wrapper script used to propagate the last directory. There are other problems with that script, and fixing all this mess is already in TODO. > I have patched it to do this (by editing mc.ext and starting mc with > "/usr/bin/env TMPDIR=~/.mc/tmp") and it works splendidly this way. Of > course the proper fix simply requires replacing TMPDIR with HOME/.mc/tmp > in the right place(s). I think that ignoring TMPDIR is a bad thing. TMPDIR is used to allow administrators to specify where all temporary files should be stored, and this directory is usually located on a very fast drive or even in memory. > Is there a reason it does not do this by default? See above. > Should I submit a proper patch? Implementation of your suggestions would not be a proper patch. -- Regards, Pavel Roskin From pehrens at ligo.caltech.edu Fri Oct 4 23:10:07 2002 From: pehrens at ligo.caltech.edu (Philip Ehrens) Date: Fri, 4 Oct 2002 16:10:07 -0700 Subject: executable scripts in /tmp Message-ID: <20021004231007.GH5715@ligo.caltech.edu> Hi Pavel, > Pavel Roskin wrote: > > Hi, Philip! > > > > > mc currently writes executable scripts into /tmp. > > > > I'll fix it. Added to src/TODO and committed. > > > > > I think this is something that should generally be avoided. > > > > I haven't heard that it should be avoided in general (apart from > > portability to 20 years old UNIX without "#!" support). Any references? It seems that there are several rootkits out there that rely on being able to execute files written to /tmp. Sorry, no specific references handy. > > > On my pokey little server I have /tmp mounted noexec because > > > I am paranoid. > > > > OK, reliance on the ability to execute scripts seems unnecessary. I agree. > > > Why, since ~/.mc/tmp exists, and is used by mc anyway, doesn't > > > mc just use ~/.mc/tmp for everything? > > > > There is no reason whatsoever to believe that the home directory is less > > likely to be mounted noexec than /tmp. Really paranoid sysadmins would > > make both noexec :-) > > > > On the other hand, I've seen installations where the home directory is > > mounted over NFS and shared between several machines. That's not a good > > choice for temporary files. Quite right. Should've thought of the NFS thing, since I beat my head against it soundly all day long at work ;^) Better to continue writing into /tmp, but no executable scripts there. Phil -- Phil Ehrens | Fun stuff: The LIGO Laboratory, MS 18-34 | http://www.ralphmag.org California Institute of Technology | http://www.yellow5.com 1200 East California Blvd. | http://www.total.net/~fishnet/ Pasadena, CA 91125 USA | http://slashdot.org Phone:(626)395-8518 Fax:(626)793-9744 | http://kame56.homepage.com From ury at bofh.homeunix.org Sat Oct 5 09:41:35 2002 From: ury at bofh.homeunix.org (Ury Stankevich) Date: Sat, 5 Oct 2002 12:41:35 +0300 Subject: ftp handling bug possible (affected 4.6.0pre1) Message-ID: <20021005124135.40a3a637.ury@bofh.homeunix.org> hi, i use mc and it's so well but... if on my disk files named are: 1992 - Scenes 1994 - Introduction then if i try access them via ftp console command - all ok, i see this list too, but when i use mc ftp connection -> i see /- Introduction /- Scenes and as result impossible enter them and access any files in this directories. __Ury From dooligan at intergate.ca Sun Oct 6 23:35:20 2002 From: dooligan at intergate.ca (Miven Dooligan) Date: Sun, 06 Oct 2002 16:35:20 -0700 Subject: viewer / quick view mode Message-ID: <3DA0C8B8.6060705@intergate.ca> Hello. I can see what you meant when you said the hex viewer is just one big hack. It looks like one of those cool after-thoughts, that kept getting more and more features ("I just need to fix two bytes! Can you make it edit?"). I've always felt that it deserves its own subdir like the editor. So, I've been beating on the viewer a bit. Some more stuff for your Hex Editor Fixit List:) If I have the right panel set to 80 columns, quick view/hex mode won't do 16 byte rows, I have to go up to 82 columns. I see that's because of the 2 columns used by the border. It bumps down to 12 byte rows even though there's still plenty of room. Perhaps we can get rid of the vertical bars separating the 4 byte groups, and it would do 16 byte rows even at 76 columns, with just a double space between 4 byte columns. Of course, then it doesn't look as good on standard 80 column displays, with a big gap between the hex and text areas, so we leave them alone unless view->widget.cols < 80. The attached patch does the job for that. hunk1: fixes bytes per line when cols < 80. hunk2: stops printing vline when cols < 80 hunk3: fixes missing space between addresses and hexdump when view->have_frame. Another mention is that the hex mode of the viewer grabs the TAB key for flipping between hex and text fields. It means you have to use your mouse to get into the other panel in quick view/hex mode, and you can't TAB back into the hex editor either or even use the mouse to get back in, so you're stuck in hex mode. The TAB key just seems to stop working. I have to exit mc and restart it to get it back. If I switch the panel to info mode, then back to hex view mode, the hex output is all on just one line, running right off the edge of the screen, or sometimes just a single column of bytes. If you do a sigwinch a few columns to where it would have to recalculate bytes_per_line anyway, it fixes itself. It seems to miss view_update_bytes_per_line() under some situations, or view->widget.cols is not set properly. I'm not at all sure where that one comes from. One other thing about the viewer: if you resize your window badly while in the viewer (either mode) or editor and return it to its original size, when you leave the viewer, the left panel has been resized (much smaller). It seems the sigwinch handler fixes the panels even when in the viewer/editor and doesn't try to return them to their original values (as per ini). It almost calls for the ability to pull the panel divider back and forth with the mouse. Cheers, Mr.Picky :) -------------- next part -------------- An embedded and charset-unspecified text was scrubbed... Name: hexedit-column-fix.patch URL: From proski at gnu.org Mon Oct 7 07:15:54 2002 From: proski at gnu.org (Pavel Roskin) Date: Mon, 7 Oct 2002 03:15:54 -0400 (EDT) Subject: Included S-Lang updated to 1.4.5 Message-ID: Hello! I have updated the S-Lang library included with the mc sources to the version 1.4.5. I actually wanted to fix a problem with the old version. When selecting text in the viewer or in the editor using mouse on xterm, the text in the clipboard had spaces after the line ends. Since Red Hat 8.0 packaged S-Lang with UTF-8 support, and there are many problems with it, I wanted to have an option that I could recommend as a temporary solution. The problem with the clipboard would not allow me to recommend the --with-screen=mcslang option to anyone. Unfortunately, the changes required to be backported were too radical. To tell the terminal where the line ends, the new S-Lang uses a new capability to erase text using current background. So I decided to update S-Lang, and it was surprisingly easy. To avoid renaming slang.h to slang-mc.h I moved all includes to a separate "include" directory. This would prevent the configure script from picking local slang/slang.h header when testing for the installed library. The new code has been tested on Red Hat 8.0 only. Please give it more testing. -- Regards, Pavel Roskin From makovick at kmlinux.fjfi.cvut.cz Mon Oct 7 08:27:55 2002 From: makovick at kmlinux.fjfi.cvut.cz (Jindrich Makovicka) Date: Mon, 07 Oct 2002 10:27:55 +0200 Subject: [PATCH] ftpfs.c linear_abort workaround for dumb servers Message-ID: <3DA1458B.7040204@kmlinux.fjfi.cvut.cz> Hello, I recently encountered some ftp servers, which keep sending data even after reporting ABOR cmd complete. This patch adds a timeout after which a reconnect is performed instead of further fetching of the remaining data. Regards, -- Jindrich Makovicka -------------- next part -------------- An embedded and charset-unspecified text was scrubbed... Name: ftpfs.diff URL: From spambox at centrum.cz Mon Oct 7 09:42:48 2002 From: spambox at centrum.cz (Jiri Mencak) Date: Mon, 7 Oct 2002 10:42:48 +0100 Subject: copy_current_pathname() In-Reply-To: References: <20021004135103.GA30648@mencak.homeip.net> Message-ID: <20021007094248.GA12426@mencak.homeip.net> On Fri, Oct 04, 2002 at 04:46:19PM -0400, Pavel Roskin wrote: > > when copying current pathname containing characters like > > [\'\\\n\t\r":; ] (and many others---for a full list see `util.c', > > name_quote()) on the MC's command line, the function name_quote() > > is not used. Is this intentional? If not, the patch for > > MC 4.6.0-pre1 is attached. > > I believe it's fixed already in the current snapshot. You are absolutely right. I'd always better check current snapshot before posting. Sorry. Jiri From proski at gnu.org Mon Oct 7 17:39:50 2002 From: proski at gnu.org (Pavel Roskin) Date: Mon, 7 Oct 2002 13:39:50 -0400 (EDT) Subject: [PATCH] ftpfs.c linear_abort workaround for dumb servers In-Reply-To: <3DA1458B.7040204@kmlinux.fjfi.cvut.cz> Message-ID: Hello! > I recently encountered some ftp servers, which keep sending data even > after reporting ABOR cmd complete. This patch adds a timeout after which > a reconnect is performed instead of further fetching of the remaining > data. How big is the timeout? Is it one second? I think it's too short. Let's give the server at least 5 seconds to abort. Could you please make a #define for the timeout and add a short comment? -- Regards, Pavel Roskin From proski at gnu.org Mon Oct 7 17:41:48 2002 From: proski at gnu.org (Pavel Roskin) Date: Mon, 7 Oct 2002 13:41:48 -0400 (EDT) Subject: Included S-Lang updated to 1.4.5 In-Reply-To: Message-ID: > So I decided to update S-Lang, and it was surprisingly easy. To avoid > renaming slang.h to slang-mc.h I moved all includes to a separate > "include" directory. This would prevent the configure script from picking > local slang/slang.h header when testing for the installed library. As a side effect, everybody using patches should now use "patch -E" to make sure that the old headers are removed. -- Regards, Pavel Roskin From proski at gnu.org Mon Oct 7 17:47:25 2002 From: proski at gnu.org (Pavel Roskin) Date: Mon, 7 Oct 2002 13:47:25 -0400 (EDT) Subject: ftp handling bug possible (affected 4.6.0pre1) In-Reply-To: <20021005124135.40a3a637.ury@bofh.homeunix.org> Message-ID: Hello! > then if i try access them via ftp console command - all ok, i see this > list too, but when i use mc ftp connection -> i see > > /- Introduction > /- Scenes That's the most reported bug in mc. I have added it to the TODO list. Help will be appreciated, but make sure that you use the CVS version or the snapshot as the starting point - there were some recent changes in the date parser. The parser function is called vfs_parse_ls_lga and is located in vfs/vfs.c. -- Regards, Pavel Roskin From proski at gnu.org Mon Oct 7 18:11:12 2002 From: proski at gnu.org (Pavel Roskin) Date: Mon, 7 Oct 2002 14:11:12 -0400 (EDT) Subject: viewer / quick view mode In-Reply-To: <3DA0C8B8.6060705@intergate.ca> Message-ID: > The attached patch does the job for that. > hunk1: fixes bytes per line when cols < 80. > hunk2: stops printing vline when cols < 80 > hunk3: fixes missing space between addresses and hexdump when > view->have_frame. Thanks for the patch and for the detailed explanation! I've applied your patch. -- Regards, Pavel Roskin From lists at pervalidus.net Mon Oct 7 23:07:00 2002 From: lists at pervalidus.net (=?iso-8859-1?B?RnLpZOlyaWMgTC4gVy4=?= Meunier) Date: Mon, 7 Oct 2002 20:07:00 -0300 Subject: Included S-Lang updated to 1.4.5 Message-ID: <20021007230700.GA8272@pervalidus> Pavel Roskin wrote: > I have updated the S-Lang library included with the mc > sources to the version 1.4.5. Stupid question, but does this also mean there are no differences between the included and a system (without UTF-8 support) 1.4.5 S-Lang ? What would you recommend if I have 1.4.5 without UTF-8 support ? --with-screen=slang ? -- 0 at pervalidus.{net, {dyndns.}org} From proski at gnu.org Mon Oct 7 23:53:09 2002 From: proski at gnu.org (Pavel Roskin) Date: Mon, 7 Oct 2002 19:53:09 -0400 (EDT) Subject: Included S-Lang updated to 1.4.5 In-Reply-To: <20021007230700.GA8272@pervalidus> Message-ID: Hello! > > I have updated the S-Lang library included with the mc > > sources to the version 1.4.5. > > Stupid question, but does this also mean there are no differences > between the included and a system (without UTF-8 support) 1.4.5 > S-Lang ? There is some difference. The included library used glib for memory allocation and replacement functions. Another difference is that you can choose between termcap and terminfo databases at the build time. Maybe even at runtime (-t option), although I didn't test it, and I know that if it's broken, nobody would report it. > What would you recommend if I have 1.4.5 without UTF-8 support ? > --with-screen=slang ? Yes. -- Regards, Pavel Roskin From dooligan at intergate.ca Wed Oct 9 06:12:58 2002 From: dooligan at intergate.ca (Miven Dooligan) Date: Tue, 08 Oct 2002 23:12:58 -0700 Subject: ALT ('-') and ALT('+') Message-ID: <3DA3C8EA.8060607@intergate.ca> Hello. A few questions: Is it reasonable to make ALT('=') do Select All as well? I use it all the time and that triple bucky gets a bit... bothersome. What about ALT('_') for invert selection? It uses the same keys as the rest of the select functions and it's a bit easier of a stretch than ALT-SHIFT-8 :) Attached is a meager patch if that seems like a good idea. * I see that those functions avoid the directories. It's a paltry matter to comment out the lines in cmd.c that check for dirs, but there must be a reason someone did that to begin with. It is a good idea when you are fixing file permissions and stuff, because you almost always want to treat the files separate from the dirs, so what if we just comment out the lines in cmd.c:reverse_selection_cmd_panel(), that way you can select all, then reverse the selection to get the directories all selected? Two keystrokes. Actually three because you have to hit ENTER after the select all. I like it, but I could see why someone else wouldn't. Usually the subdirs are a minor component of a directory structure, and it's the files you want to work on, usually. Also, the unselect-all SHOULD do directories, shouldn't it. That's what it's for, to unselect everything, isn't it? Right now it won't touch selected directories. Isn't that wrong? If it's not wrong, could someone please explain why? Maybe it's because this is a FILE manager, not a DIRECTORY manager? :) :) LOL... Anyway, attached are a couple minor patches for those things if they seem like the right thing to do. Cheers, Mr.Picky :) -------------- next part -------------- An embedded and charset-unspecified text was scrubbed... Name: select-all-key.patch URL: -------------- next part -------------- An embedded and charset-unspecified text was scrubbed... Name: invert-selection.patch URL: -------------- next part -------------- An embedded and charset-unspecified text was scrubbed... Name: unselect-all.patch URL: From proski at gnu.org Wed Oct 9 08:27:31 2002 From: proski at gnu.org (Pavel Roskin) Date: Wed, 9 Oct 2002 04:27:31 -0400 (EDT) Subject: ALT ('-') and ALT('+') In-Reply-To: <3DA3C8EA.8060607@intergate.ca> Message-ID: Hello! > Is it reasonable to make ALT('=') do Select All as well? I use it all > the time and that triple bucky gets a bit... bothersome. The code for the file selection keys is messy and needs a serious cleanup. Adding new key combination to compensate for brokeness of the existing code is hardly reasonable. Additional complexity will be an additional obstacle for those who will clean up this code. People learn keys, and then you can reassign those keys to something better. There are several things you can do to make it more convenient for now. You can use a terminal that has separate escape sequences for the keypad. You can use only_leading_plus_minus. You can even use "Learn keys" to assign your favorite combinations. Have you tried "Learn keys" for that? If yes, why didn't you like this approach? I think that the "Learn keys" dialog has two problems - one is that users cannot find it when they need it. The other is that they don't realize that it can help them. The answer (hard, but still not ideal) would be to implement remappable keys and a to write a much more complex dialog to assign keys to actions (as opposed to matching physical keystrokes with escape sequences). This would also mean stripping hardcoded keys from the menus, breaking all translations. It was discussed many times, but nothing was done. I think it can be said about the current state of the project that most of the simple patches have been applied. The remaining stuff is mostly hard and involves redesign. The unfortunate thing is that very few people actually want to do the hard stuff. Most contibutors simply try to "overstretch" old bad design. Remember e.g. chaining video players with the "||" operator, which was an obvious (and wrong) extension of the same idea for the Word doc viewers. > I see that those functions avoid the directories. It's a paltry matter > to comment out the lines in cmd.c that check for dirs, but there must be > a reason someone did that to begin with. At least two reasons. One is that the new dialogs are harder to code than to use input_dialog() or other ready dialog. Just a little details - when XView was supported, all the dialog had to be created backwards, i.e. starting with the widgets that are last in the Tab order. Even the title of the dialog had to be drawn manually until recently. Most importantly - the dialog API is undocumented. The other reason is that the selection for directories is in fact supported. Just add slash at the end of the pattern. Usability of this feature was obviously neglected, so that not only most users are unaware of it, but even the developers trying to improve this code! I cannot apply your patch for two reasons: 1) It makes it harder to select only files if I want to. 2) It replaces one random choice with another one instead of giving the user the choice (and giving the choice to the user means a checkbox with an obvious name, not an extra paragraph in the manual). > that way you can select all, then reverse the selection to get the > directories all selected? Two keystrokes. Don't you realize that this inconsistency (invert affects directories, select doesn't) is just another bug? It's not right, it's just tolerated by users, unlike files on ftp beginning with "2000 " :-) > Usually the subdirs are a minor component of a directory structure, and > it's the files you want to work on, usually. Let the user decide. I'll appreciate if you check popular non-free file managers (Far, Norton Commander, Windows Commander) and make the behaviour consistent with at least with one of them if possible. At least some users won't have to learn news keys when they start using mc. > Also, the unselect-all SHOULD do directories, shouldn't it. That's what > it's for, to unselect everything, isn't it? Right now it won't touch > selected directories. Isn't that wrong? If it's not wrong, could > someone please explain why? Maybe because something (just like like you) didn't like writing new dialogs? -- Regards, Pavel Roskin From Rach.Holmes at GMX.net Wed Oct 9 15:06:06 2002 From: Rach.Holmes at GMX.net (Rach Holmes) Date: Wed, 9 Oct 2002 16:06:06 +0100 Subject: Circular referencing macro segfault Message-ID: <20021009160606.298cf062.Rach.Holmes@GMX.net> Hi, There is a macro circular referencing bug in mcedit from version 4.5.55, 4.6.0pre1 and the current snapshot I am using(mc-2002-10-08-21 from ibiblio) and possibly other versions. I have seen no data corruption occur, and the "fix" so far has been to think a little more carefully while assigning macro keys :) An example to replicate the segfault: C-r C-a C-r-a # Setup a macro on 'a' calling the macro (defined or not) on 'a' to set everything up, then simply: C-a-a to see the segfault. I have tested it with the versions noted above, and also I have tested on different machines with different configurations. A back trace has been left out because this is not an intermittent error. It will still occur if the loop is bigger, ie. macro a->b->c->a (this was the situation when it showed up). Sometimes it requires reset(1) afterwards, other times it doesn't. Rach From proski at gnu.org Wed Oct 9 15:32:11 2002 From: proski at gnu.org (Pavel Roskin) Date: Wed, 9 Oct 2002 11:32:11 -0400 (EDT) Subject: Circular referencing macro segfault In-Reply-To: <20021009160606.298cf062.Rach.Holmes@GMX.net> Message-ID: Hello! > It will still occur if the loop is bigger, ie. macro a->b->c->a (this > was the situation when it showed up). Sometimes it requires reset(1) > afterwards, other times it doesn't. You forgot the attachment. I don't think I'll apply it, because you are the first person who reports this bug. There is no urgency to apply a temporary fix that only fixes the simplest loops. -- Regards, Pavel Roskin From dooligan at intergate.ca Wed Oct 9 21:10:14 2002 From: dooligan at intergate.ca (Miven Dooligan) Date: Wed, 09 Oct 2002 14:10:14 -0700 Subject: ALT ('-') and ALT('+') References: Message-ID: <3DA49B36.8040800@intergate.ca> > > > >Have you tried "Learn keys" for that? If yes, why didn't you like this >approach? I think that the "Learn keys" dialog has two problems - one is >that users cannot find it when they need it. The other is that they don't >realize that it can help them. > Yes, it's oriented around the individual keys rather than the functions associated with them, and it's not necessarily obvious that you can re-assign the functions this way. I messed with it once, and didn't figure it could help me tweak keystroke combos like that. Maybe that's a dialog that needs a bunch of work. > >The other reason is that the selection for directories is in fact >supported. Just add slash at the end of the pattern. Usability of this >feature was obviously neglected, so that not only most users are unaware >of it, but even the developers trying to improve this code! > Yes, it's true. I had no idea. That's great though. Y'see how important your experience is? It was the directory thing that was bothering me, and there is already a (little known) feature that does what I want. I thought there must be some reason someone hadn't already done this. What about a small patch to add that info to the select/unselect dialogs: "Append '/' for directories". (patch attached :) As you say, there's many users who don't know that. >>Also, the unselect-all SHOULD do directories, shouldn't it. That's what >>it's for, to unselect everything, isn't it? Right now it won't touch >>selected directories. Isn't that wrong? If it's not wrong, could >>someone please explain why? >> > >Maybe because something (just like like you) didn't like writing new >dialogs? > Heh heh. I like to check in here first to see if its worth the time and effort, or I'm just missing something (usually the case:). Cheers. -------------- next part -------------- An embedded and charset-unspecified text was scrubbed... Name: select-dialog-dir.patch URL: From proski at gnu.org Thu Oct 10 16:59:12 2002 From: proski at gnu.org (Pavel Roskin) Date: Thu, 10 Oct 2002 12:59:12 -0400 (EDT) Subject: ALT ('-') and ALT('+') In-Reply-To: <3DA49B36.8040800@intergate.ca> Message-ID: Hello! > What about a small patch to add that info to the select/unselect > dialogs: "Append '/' for directories". (patch attached :) You made that message non-translatable, but I guarantee you that somebody will make it translarable very soon if this patch is applied. It's just odd to have part of the dialog in English and part, say, in French. I must say that I really don't like the fact that you didn't comment why you made this message non-translatable. I don't like attempts to hide hacks. The amount of time spent on testing a patch depends on the trust to the author of that patch. Please think about it. All user-visible texts are translated into 30 languages. It takes time of many people. I think we should do it right the first time rather than resort to quick hacks when it comes to the user interface. > Heh heh. I like to check in here first to see if its worth the time and > effort, or I'm just missing something (usually the case:). Yes, that's the right approach. It is worth the time and effort if you do it right. -- Regards, Pavel Roskin From shyong88 at yahoo.com.tw Fri Oct 11 04:27:58 2002 From: shyong88 at yahoo.com.tw (=?big5?B?qV2/37Ha?=) Date: Fri, 11 Oct 2002 12:27:58 +0800 Subject: Midnight Commander Message-ID: <000f01c270de$9e0fa030$0301a8c0@SUM> ??Excuse me , I don't know how to use the software of mc-4.6.0-pre1.tar.gz with install to FreeBsd 4.6.2 . So , could you talk with me about how to install the Midnight Commander , i don't know to install steps and please talk me , thank you .... Come from Taiwan user From dooligan at intergate.ca Sat Oct 12 00:09:47 2002 From: dooligan at intergate.ca (Miven Dooligan) Date: Fri, 11 Oct 2002 17:09:47 -0700 Subject: Select box message translation? References: Message-ID: <3DA7684B.4010805@intergate.ca> Hello. > >You made that message non-translatable, but I guarantee you that somebody >will make it translarable very soon if this patch is applied. It's just >odd to have part of the dialog in English and part, say, in French. > Translatable? OK, you got me there. :) I've looked at those .Po files, and short of manually pasting the line into each one, is there something else I can do? Should I add it to the French one ("Ajouter \"/\" pour les repertoires.") and let others take it from there? How does .po work? I've never touched it :( being a spoiled anglo. Is there a HOWTO-make-stuff-translatable.html somewhere? Cheers. From pehrens at ligo.caltech.edu Sat Oct 12 00:07:03 2002 From: pehrens at ligo.caltech.edu (Philip Ehrens) Date: Fri, 11 Oct 2002 17:07:03 -0700 Subject: .po potfiles In-Reply-To: <3DA7684B.4010805@intergate.ca> References: <3DA7684B.4010805@intergate.ca> Message-ID: <20021012000703.GA9721@ligo.caltech.edu> See the docs for gnu gettext, whitey: http://www.delorie.com/gnu/docs/gettext/gettext_toc.html Miven Dooligan wrote: > > I've looked at those .Po files, and short of manually > pasting the line into each one, is there something else > I can do? Should I add it to the French one > ("Ajouter \"/\" pour les repertoires.") and let others > take it from there? How does .po work? I've never > touched it :( being a spoiled anglo. From proski at gnu.org Sat Oct 12 00:16:27 2002 From: proski at gnu.org (Pavel Roskin) Date: Fri, 11 Oct 2002 20:16:27 -0400 (EDT) Subject: Select box message translation? In-Reply-To: <3DA7684B.4010805@intergate.ca> Message-ID: Hello! > >You made that message non-translatable, but I guarantee you that somebody > >will make it translarable very soon if this patch is applied. It's just > >odd to have part of the dialog in English and part, say, in French. > > > Translatable? OK, you got me there. :) Good. I actually wasn't sure whether you did it intentionally, so that the translators would not waste their time on the temporary hack until somebody (i.e. me) does it right. That's why I said that I don't like it. > I've looked at those .Po files, and short of manually pasting the line > into each one, is there something else I can do? Should I add it to the > French one ("Ajouter \"/\" pour les repertoires.") and let others take > it from there? How does .po work? I've never touched it :( being a > spoiled anglo. There was a translatable string on the same line you changed! Just replace "String" with _("String") and it becomes translatable. That's it. Translations are updated using special make targets in the "po" directory, and you should worry about them unless you want to help with some of the translations. > Is there a HOWTO-make-stuff-translatable.html somewhere? info gettext Google sometimes helps too. -- Regards, Pavel Roskin From proski at gnu.org Sat Oct 12 00:19:38 2002 From: proski at gnu.org (Pavel Roskin) Date: Fri, 11 Oct 2002 20:19:38 -0400 (EDT) Subject: Midnight Commander In-Reply-To: <000f01c270de$9e0fa030$0301a8c0@SUM> Message-ID: Hello! On Fri, 11 Oct 2002, [big5] ?]???? wrote: > ???MExcuse me , I don't know how to use the software of mc-4.6.0-pre1.tar.gz > with install to FreeBsd 4.6.2 . > > So , could you talk with me about how to install the Midnight Commander , i > don't know to install steps and > > please talk me , thank you .... The process is described in the file called "INSTALL". Follow the steps described there. If you have any problems, report them to mc at gnome.org, not to this list. -- Regards, Pavel Roskin From dmartina at excite.com Mon Oct 14 09:47:24 2002 From: dmartina at excite.com (David Martin) Date: Mon, 14 Oct 2002 05:47:24 -0400 (EDT) Subject: Spanish translations & Mandrake 9.0 report Message-ID: <20021014094724.477D23E13@xmxpita.excite.com> Here are some changes and updates for the Spanish help, messages and hint files. There are some tweaks for the English stuff, check them. A sentence about termination in the editor was recently removed in mc.1.in. I think it was just a message to tell about the autoindent option available when pressing return and it had nothing to do with termination. I rewrote and reordered that piece in the Spanish help. There are some cosmetic changes about double vertical spacing in some pages. It would be nice to have a uniform appearance. I'm thinking of printing the manual page and see what does it look like. I was quite busy this week but I installed and I tried mc with Mandrake 9.0. It includes links 2.1pre2 and it seems that the -dump option has been removed. Sniff. You get a warning about it when viewing HTML files and I'm afraid that lynx is used instead. Is it possible to have some checking for manpages location? Mandrake has its man pages stored in /usr/share/man. I use --prefix=/usr when compiling mc and I have to issue --mandir=/usr/share/man to get the manual there instead of /usr/man, which in turn doesn't exist and it's created by mc when you make install. I'm getting a warning about duplicated definitions for something about the mouse. I also get a warning about libcom_err versions. I'm including the whole make.log compressed (10K, not too long I hope). I'm talking about latest snapshots, you know. I still have to fine tune the installation and check how does the wheel mouse behave; I got some messages about some liblow.c but I think it has to do with my laptop and the external and internal PS/2 mouse. When updating the .po files I'm using make update-po and gettext 0.11.2 and it seems to go fine. I used to do this with the update.pl and update.sh from CVS (old versions as I haven't used cvs for ages), but now I need to remove the --add-comments for getttext tools used inside the script. It's true that .po files weren't updated for snapshots and I was patching my own previous file. That's all I have for now. *8-) David ------------------------------------------------ Join Excite! - http://www.excite.com The most personalized portal on the Web! -------------- next part -------------- An HTML attachment was scrubbed... URL: -------------- next part -------------- A non-text attachment was scrubbed... Name: pending15.diff.bz2 Type: application/octet-stream Size: 8468 bytes Desc: not available URL: -------------- next part -------------- A non-text attachment was scrubbed... Name: make.log.bz2 Type: application/octet-stream Size: 4933 bytes Desc: not available URL: From proski at gnu.org Mon Oct 14 21:46:51 2002 From: proski at gnu.org (Pavel Roskin) Date: Mon, 14 Oct 2002 17:46:51 -0400 (EDT) Subject: Spanish translations & Mandrake 9.0 report In-Reply-To: <20021014094724.477D23E13@xmxpita.excite.com> Message-ID: Hi, David! > Here are some changes and updates for the Spanish help, messages and > hint files. There are some tweaks for the English stuff, check them. Thanks, applied. > A sentence about termination in the editor was recently removed in > mc.1.in. I think it was just a message to tell about the autoindent > option available when pressing return and it had nothing to do with > termination. I rewrote and reordered that piece in the Spanish help. I have no idea what it means. mc.1.in didn't have the word "termination" one month ago, according to CVS. > There are some cosmetic changes about double vertical spacing in some > pages. It would be nice to have a uniform appearance. I'm thinking of > printing the manual page and see what does it look like. I really appreciate that you are working on uniform appearance and professionally looking documentation. I wrote you in private, but I'll repeat for the list - I want you to have write access to CVS and to be a member of the development team. > I was quite busy this week but I installed and I tried mc with Mandrake > 9.0. It includes links 2.1pre2 and it seems that the -dump option has > been removed. Sniff. You get a warning about it when viewing HTML files > and I'm afraid that lynx is used instead. That's my error. Fixed in CVS. links never used "-dump". > Is it possible to have some checking for manpages location? Mandrake has > its man pages stored in /usr/share/man. I use --prefix=/usr when > compiling mc and I have to issue --mandir=/usr/share/man to get the > manual there instead of /usr/man, which in turn doesn't exist and it's > created by mc when you make install. GNU Coding Standards differ from Linux Standard Base on this. Autoconf follows GNU Coding Standards, quite unsurprisingly. I think we can change the default for mc, but I believe it doesn't solve the underlying problem. You can make a symlink or fix man to search both paths. You can also ask GNU and LSB guys to sit together and decide on the issue, but please make sure they didn't do it already. > I'm getting a warning about duplicated definitions for something about > the mouse. It looks like gpm in Mandrake uses the same symbols for the wheel buttons as mc. I'll have a look. > I also get a warning about libcom_err versions. Perhaps Mandrake compiled e2fsutils against old libraries? I don't know. It's hardly related to mc. Report to the Mandrake maintainer of e2fsutils. > I'm including the whole make.log compressed (10K, not too long I hope). > I'm talking about latest snapshots, you know. I still have to fine tune > the installation and check how does the wheel mouse behave; I got some > messages about some liblow.c but I think it has to do with my laptop and > the external and internal PS/2 mouse. liblow.c is from gpm. I guess mc is not quite compatible with the latest gpm. It's pretty depressing. I'm spending so much time working on mc, yet the project becomes more and more obsolete. It has problems with Mandrake gpm, with Red Hat S-Lang, it uses manpages with those bizarre groff macros, that even GNU calls obsolete. My only hope that the cleanup we are doing will make it easier for the new contributors to join, although the new generation seems to prefer GUI. > When updating the .po files I'm using make update-po and gettext 0.11.2 > and it seems to go fine. I used to do this with the update.pl and > update.sh from CVS (old versions as I haven't used cvs for ages), but > now I need to remove the --add-comments for getttext tools used inside > the script. It's true that .po files weren't updated for snapshots and I > was patching my own previous file. Actually, it should be --add-comments=TRANSLATORS:, so that only comments starting with "TRANSLATORS:" appear in the po-files. That's what the Makefile installed by gettext 0.11.5 does. Fixed in CVS. Thank you! -- Regards, Pavel Roskin From dmartina at excite.com Tue Oct 15 10:38:26 2002 From: dmartina at excite.com (David Martin) Date: Tue, 15 Oct 2002 06:38:26 -0400 (EDT) Subject: Spanish translations & Mandrake 9.0 report Message-ID: <20021015103826.7DC503DE0@xmxpita.excite.com> Sorry, it was "completion", not "termination". http://cvs.gnome.org/bonsai/cvsview2.cgi?diff_mode=context&whitespace_mode=show&file=mc.1.in&branch=&root=/cvs/gnome&subdir=mc/doc&command=DIFF_FRAMESET&rev1=1.62&rev2=1.63 "The completion key also does a Return with an automatic indent." --- On Mon 10/14, Pavel Roskin wrote: > > A sentence about termination in the editor was recently removed in > > mc.1.in. I think it was just a message to tell about the autoindent > > option available when pressing return and it had nothing to do with > > termination. I rewrote and reordered that piece in the Spanish help. > > I have no idea what it means. mc.1.in didn't have the word > "termination" > one month ago, according to CVS. > ------------------------------------------------ Join Excite! - http://www.excite.com The most personalized portal on the Web! From ury at bofh.homeunix.org Tue Oct 15 18:21:42 2002 From: ury at bofh.homeunix.org (Ury N.Stankevich) Date: Tue, 15 Oct 2002 21:21:42 +0300 Subject: [BUG] mc edit Floating point exception 4.6.0pre1 Message-ID: <20021015212142.4438e651.ury@bofh.homeunix.org> this steps cause open any file in internal editor to edit set tabulation step to 0 press and one more... i have trouble in use ftp capabilities under user: when i connect to ftp server - i see notning on panel, and in tips string i see 'ftpfs: failed; nowhere to fallback to' while under r00t - all ok. this my mistake or bug? __Ury From proski at gnu.org Tue Oct 15 19:21:41 2002 From: proski at gnu.org (Pavel Roskin) Date: Tue, 15 Oct 2002 15:21:41 -0400 (EDT) Subject: [BUG] mc edit Floating point exception 4.6.0pre1 In-Reply-To: <20021015212142.4438e651.ury@bofh.homeunix.org> Message-ID: Hello! > this steps cause > > open any file in internal editor to edit > set tabulation step to 0 > press I confirm it. > and one more... > i have trouble in use ftp capabilities under user: > when i connect to ftp server - i see notning on panel, and in tips > string i see 'ftpfs: failed; nowhere to fallback to' > while under r00t - all ok. > this my mistake or bug? There are too little details to reproduce this problem. I remember similar bugreports in the past. In any case it cannot be your mistake. I also remember seeing this message myself if the ftp server disconnects on timeout and I press Ctrl-R to reload the listing. The message is so strange because ftpfs tries to use UNIX options for "ls" and then falls back to more portable options. If that fails as well, this message is printed. In your case it makes no sense because you don't know and don't have to know what ftpfs tries to do before it fails. It's a reminder for the contributors - try to see the whole picture, not just the problem your patch tries to solve. I recommend that whoever submits a patch that changes user-visible strings also provides the translation to his/her native language. "nowhere to fallback to" may look very solid when written in English, but try writing this in your language and think again if your friends will understand what it is about when they see this message on the screen. -- Regards, Pavel Roskin From lists at pervalidus.net Tue Oct 15 22:25:26 2002 From: lists at pervalidus.net (=?iso-8859-1?B?RnLpZOlyaWMgTC4gVy4=?= Meunier) Date: Tue, 15 Oct 2002 19:25:26 -0300 Subject: slang/README should be updated or rewritten Message-ID: <20021015222526.GB4821@pervalidus.net> I think slang/README should be updated or rewritten. Something like: This is the S-Lang library, version 1.4.5. It has been modified a little to make it work with the Midnight Commander. The complete distribution can be downloaded from: http://space.mit.edu/~davis/slang/ ftp://ftp.jedsoft.org/pub/davis/slang/ -- 0 at pervalidus.{net, {dyndns.}org} From proski at gnu.org Wed Oct 16 00:07:13 2002 From: proski at gnu.org (Pavel Roskin) Date: Tue, 15 Oct 2002 20:07:13 -0400 (EDT) Subject: slang/README should be updated or rewritten In-Reply-To: <20021015222526.GB4821@pervalidus.net> Message-ID: Hello! > I think slang/README should be updated or rewritten. Something > like: Thanks. I have changed it even more and committed. -- Regards, Pavel Roskin From milamber at vakna.cz Wed Oct 16 06:35:10 2002 From: milamber at vakna.cz (=?iso-8859-2?q?Ji=F8=ED=20Hubka?=) Date: Wed, 16 Oct 2002 08:35:10 +0200 Subject: UTF8 Message-ID: <200210160824.09994.milamber@vakna.cz> Hi, I have a question. Do you prepare mc with support utf8 ? Thanks Jiri Hubka From proski at gnu.org Wed Oct 16 18:41:40 2002 From: proski at gnu.org (Pavel Roskin) Date: Wed, 16 Oct 2002 14:41:40 -0400 Subject: gnu.org is down Message-ID: <1034793700.3dadb2e4b6bd6@webmail.spamcop.net> An embedded and charset-unspecified text was scrubbed... Name: not available URL: From kai at cmail.ru Thu Oct 17 07:09:04 2002 From: kai at cmail.ru (Andrew V. Samoilov) Date: Thu, 17 Oct 2002 10:09:04 +0300 Subject: Using poptPrintHelp() References: Message-ID: <3DAE6210.10106@cmail.ru> Pavel Roskin wrote: > Hello! > > >>I suggest that we change print_mc_usage() to use poptPrintHelp(). This is >>a standard popt function from src/popthelp.c, and it's already linked into >>the mc executable. > Now it is in stream to use poptSetOtherOptionHelp() in print_mc_usage(), but this change requires massive po/*.po updates. - Regards, Andrew V. Samoilov -------------- next part -------------- An embedded and charset-unspecified text was scrubbed... Name: main.c.patch URL: From dmartina at excite.com Thu Oct 17 11:42:35 2002 From: dmartina at excite.com (David Martin) Date: Thu, 17 Oct 2002 07:42:35 -0400 (EDT) Subject: Links -dump as HTML viewer Message-ID: <20021017114235.938AC3E11@xmxpita.excite.com> --- On Mon 10/14, Pavel Roskin wrote: > > > 9.0. It includes links 2.1pre2 and it seems that the -dump option > has > > been removed. Sniff. You get a warning about it when viewing HTML > files > > and I'm afraid that lynx is used instead. > > That's my error. Fixed in CVS. links never used "-dump". Are you sure? I tried the snapshot and it didn't work. We need the -dump option to avoid entering the interactive ncurses full screen mode and getting the formatted thing to stdout. It was there in the 0.96 version but it seems to have dissappeared in this sort of 2.1pre version of Links. I was checking their Changelog files and couldn't find any notice about this. We may have to give up and use lynx again. Any idea? Any contact in the links development team? It's a bit unfair to remove such feature... *8-) David P.S.: I'll sort the CVS thing ASAP. _______________________________________________ Join Excite! - http://www.excite.com The most personalized portal on the Web! -------------- next part -------------- An HTML attachment was scrubbed... URL: From maciek at alpha.net.pl Thu Oct 17 10:19:19 2002 From: maciek at alpha.net.pl (Maciek Kurkiewicz) Date: Thu, 17 Oct 2002 12:19:19 +0200 Subject: mc bashrc Message-ID: <000f01c275c6$a9e2c260$0601a8c0@kurkiewicz> I want to write some scripts in bashrc fille, my friend has it in ~/.mc/bashrc in mc version 4.5.54 and in his man mc is: "If you are using bash you can specify startup commands for the subshell in your ~/.mc/bashrc file and special keyboard maps in the ~/.mc/inputrc file. tcsh users may specify startup commands in the ~/.mc/tcshrc file" I have wrote scrpit in ~/.bashrc script -> it works like that : checking who is logged as a root if it is logged from my computer, he donn't write history in ~/.bash_history if logged from other comuters as a root it is allowed to write history But ... when I start `mc` and perss CTRL+O and write some commands and this commands are remembered in ~/.bash_history. It is possibility to write script in ~/.mc/bashrc like in ~/.bashrc but i don't have that file in mc. How can I do it ? best wishes from Poland (sorry for mistakes) Maciek Kurkiewicz maciek at alpha.net.pl P.S. I have: GNU Midnight Commander 4.5.55 Mandrake 8.2 -> polish language -- ---- Oferta jakiej jeszcze nie by?o! Serwer www 60 MB za 99 z? rocznie Szczeg??y: www.oferta.alpha.pl ---- From kai at cmail.ru Thu Oct 17 15:44:24 2002 From: kai at cmail.ru (Andrew V. Samoilov) Date: Thu, 17 Oct 2002 18:44:24 +0300 Subject: man2hlp handles .TP and .IP without arguments Message-ID: <3DAEDAD8.2080904@cmail.ru> Hello, we are one more step toward to make xnc.hlp a text file. .IP and .TP paragraphs are how processed by man2hlp, but their arguments are ignored now. Now we need to implement .OL (ordered list) from nroff mm package and possible .RS and .RE from man one. -- Regards, Andrew V. Samoilov -------------- next part -------------- An embedded and charset-unspecified text was scrubbed... Name: man2hlp.c.patch URL: From nerijus at users.sourceforge.net Thu Oct 17 17:13:14 2002 From: nerijus at users.sourceforge.net (Nerijus Baliunas) Date: Thu, 17 Oct 2002 19:13:14 +0200 (EET) Subject: Links -dump as HTML viewer In-Reply-To: <20021017114235.938AC3E11@xmxpita.excite.com> References: <20021017114235.938AC3E11@xmxpita.excite.com> Message-ID: <200210171717.g9HHHwPZ023093@mx.ktv.lt> On Thu, 17 Oct 2002 07:42:35 -0400 (EDT) David Martin wrote: > > > 9.0. It includes links 2.1pre2 and it seems that the -dump option has > > > been removed. Sniff. You get a warning about it when viewing HTML files > > > and I'm afraid that lynx is used instead. > > > > That's my error. Fixed in CVS. links never used "-dump". > > Are you sure? I tried the snapshot and it didn't work. We need the -dump option to avoid entering the interactive ncurses full screen mode and getting the formatted thing to stdout. > > It was there in the 0.96 version but it seems to have dissappeared in this sort of 2.1pre version of Links. I was checking their Changelog files and couldn't find any notice about this. (Could you please wrap your lines at about 75 column?) links-0.96-2 from RH 7.2 supports -dump according to links -help. RH 7.3 includes elinks-0.3.0-1 and RH 8.0 includes elinks-0.3.2-1 (http://elinks.or.cz/; executable is named 'links'). Both support -dump. I'd suggest leaving -dump and probably talking to links developers, if this option is removed in newer links version. Regards, Nerijus From proski at gnu.org Fri Oct 18 22:56:17 2002 From: proski at gnu.org (Pavel Roskin) Date: Fri, 18 Oct 2002 18:56:17 -0400 (EDT) Subject: UTF8 In-Reply-To: <200210160824.09994.milamber@vakna.cz> Message-ID: Hello! > Hi, I have a question. Do you prepare mc with support utf8 ? As far as I know, nobody is working on it. You can find some discussion here: http://mail.gnome.org/archives/mc/2002-October/msg00065.html -- Regards, Pavel Roskin From proski at gnu.org Mon Oct 21 00:47:38 2002 From: proski at gnu.org (Pavel Roskin) Date: Sun, 20 Oct 2002 20:47:38 -0400 (EDT) Subject: mc bashrc In-Reply-To: <000f01c275c6$a9e2c260$0601a8c0@kurkiewicz> Message-ID: Hello! > But ... when I start `mc` and perss CTRL+O and write some commands and > this commands are remembered in ~/.bash_history. It is possibility to > write script in ~/.mc/bashrc like in ~/.bashrc but i don't have that > file in mc. How can I do it ? I think you are very unclear about your problem. Please try a formal description: 1) What you do 2) What you expect to get 3) What you actually get I just checked if .mc/bashrc actually works. Yes, it works for me. -- Regards, Pavel Roskin From proski at gnu.org Mon Oct 21 02:15:59 2002 From: proski at gnu.org (Pavel Roskin) Date: Sun, 20 Oct 2002 22:15:59 -0400 (EDT) Subject: Circular referencing macro segfault In-Reply-To: <20021009160606.298cf062.Rach.Holmes@GMX.net> Message-ID: Hello! > There is a macro circular referencing bug in mcedit from version > 4.5.55, 4.6.0pre1 and the current snapshot I am using(mc-2002-10-08-21 > from ibiblio) and possibly other versions. I have seen no data > corruption occur, and the "fix" so far has been to think a little more > carefully while assigning macro keys :) Fixed. It was really easy to fix - I simply added a counter for the recursion depth, and it's not allowed to go over 256. 256 recursion levels should be enough for everyone :-) > I have tested it with the versions noted above, and also I have tested > on different machines with different configurations. A back trace has > been left out because this is not an intermittent error. > > It will still occur if the loop is bigger, ie. macro a->b->c->a (this > was the situation when it showed up). Sometimes it requires reset(1) > afterwards, other times it doesn't. Sorry, I misread your message. The word "still" made me think that you were going to send a fix that only fixes the simplest kind of recursion. I hate temporary hacks, so I slightly overreacted. Sorry for that. -- Regards, Pavel Roskin From proski at gnu.org Mon Oct 21 06:37:08 2002 From: proski at gnu.org (Pavel Roskin) Date: Mon, 21 Oct 2002 02:37:08 -0400 (EDT) Subject: "Safe delete" featured in Interface Hall of Shame Message-ID: Hello! I have found this site very useful for developers of user interfaces: http://iarchitect.com/ "Safe delete" in mc: http://www.red-bean.com/~proski/safe_delete.png is strikingly similar to this: http://iarchitect.com/errormsg.htm#ERROR39 Fortunately, "Safe delete" is disabled by default. But even code disabled by default needs to be maintained, or it can become very "unsafe" to use. I have found that some translations (Russian and Japanese - just two of the four I checked) ask to type "yes" in English but compare the resulting text to the translated version of "yes". In other words, users have to guess it, or the recursive delete won't work. I have never seen bugreports about this. Does it mean that nobody uses "Safe delete" and it's safe to delete? :-) As an intermediate solution, I suggest that "Safe delete" stays, but its only effect will be changing the default button to "no" in the delete dialogs. -- Regards, Pavel Roskin From dmartina at excite.com Mon Oct 21 10:00:25 2002 From: dmartina at excite.com (David Martin) Date: Mon, 21 Oct 2002 06:00:25 -0400 (EDT) Subject: links in man2hlp Message-ID: <20021021100025.E2FD8B70B@xmxpita.excite.com> I tried this to get punctuation marks out the link references avoiding extra spaces (ala .BR or .IR). The best idea I got was using "\&" as separator because it seems to have no side effects for groff in the man page. I read something about "\&" preventing kerning between characters but that doesn't seem anything to worry about. I invented a LINK3 entry but if you trust the code it could go as LINK2. I was careful with strings but I don't know if glib functions must be used anyway. Is there anyway to get the date into the help/groff files? I (me again :) wanted to complete the header and the footer (in .TH at the head of the file). @VERSION@ seems to do the job for the version (like @prefix@) in mc.1.in. *8-) David ------------------ --- /man2hlp.c 2002-10-18 04:22:34.000000000 +0200 +++ man2hlp.c 2002-10-20 17:19:55.000000000 +0200 @@ -482,6 +482,9 @@ } else if (strcmp (buffer, ".\\\"LINK2\"") == 0) { /* Next two input lines form a link */ link_flag = 2; + } else if (strcmp (buffer, ".\\\"LINK3\"") == 0) { + /* Next two input lines form a link */ + link_flag = 4; } else if ((strcmp (buffer, ".PP") == 0) || (strcmp (buffer, ".P") == 0) || (strcmp (buffer, ".LP") == 0)) { indentation = 0; @@ -591,7 +594,9 @@ static void handle_link (char *buffer) { - static char old[80]; + static char old[BUFFER_SIZE]; + static char rest[BUFFER_SIZE]; + char *endlink; int len; switch (link_flag) { @@ -601,9 +606,31 @@ case 2: /* First part of new format link */ /* Bold text or italics text */ + + /* FIXME: Take care of .XX as .IR, .BR, etc. */ + if (buffer[0] == '.' && (buffer[1] == 'I' || buffer[1] == 'B')) + for (buffer += 2; *buffer == ' ' || *buffer == '\t'; buffer++); + snprintf (old, BUFFER_SIZE, "%s", buffer); + strcpy (rest, ""); + link_flag = 3; + break; + case 4: + /* Same as previous but using a mark to end link */ + + /* FIXME: Take care of .XX as .IR, .BR, etc. */ if (buffer[0] == '.' && (buffer[1] == 'I' || buffer[1] == 'B')) for (buffer += 2; *buffer == ' ' || *buffer == '\t'; buffer++); - strcpy (old, buffer); + + endlink = strstr(buffer,"\\&"); + + if (endlink) { + snprintf(old, BUFFER_SIZE, "%.*s", endlink-buffer, buffer); + snprintf(rest, BUFFER_SIZE, "%s", endlink+2); + } else { + snprintf (old, BUFFER_SIZE, "%s", buffer); + strcpy (rest, ""); + } + link_flag = 3; break; case 3: @@ -618,8 +645,8 @@ if (len && buffer[len - 1] == '"') { buffer[--len] = 0; } - printf_string ("%c%s%c%s%c\n", CHAR_LINK_START, old, - CHAR_LINK_POINTER, buffer, CHAR_LINK_END); + printf_string ("%c%s%c%s%c%s\n", CHAR_LINK_START, old, + CHAR_LINK_POINTER, buffer, CHAR_LINK_END, rest); link_flag = 0; /* Add to the linked list */ if (current_link) { _______________________________________________ Join Excite! - http://www.excite.com The most personalized portal on the Web! From kai at cmail.ru Mon Oct 21 11:08:21 2002 From: kai at cmail.ru (Andrew V. Samoilov) Date: Mon, 21 Oct 2002 14:08:21 +0300 Subject: links in man2hlp References: <20021021100025.E2FD8B70B@xmxpita.excite.com> Message-ID: <3DB3E025.103@cmail.ru> David Martin wrote: > > I tried this to get punctuation marks out the link references > avoiding extra spaces (ala .BR or .IR). The best idea I got > was using "\&" as separator because it seems to have no side > effects for groff in the man page. I read something about > "\&" preventing kerning between characters but that doesn't > seem anything to worry about. Another possible solution is .\" at the same line in .\"LINK2 > > I invented a LINK3 entry but if you trust the code it could go > as LINK2. I was careful with strings but I don't know if > glib functions must be used anyway. snprintf() is not portable, so use g_snprintf(). > > Is there anyway to get the date into the help/groff files? > I (me again :) wanted to complete the header and the footer > (in .TH at the head of the file). @VERSION@ seems to do the > job for the version (like @prefix@) in mc.1.in. > There are yr (years since 1900), mo (month) and dy (day of month)registers in groff. So, you can add something like .nr yr \n(yr+1900 .TH ... "\n(dy-\n(mo-\n(yr" at the beginning of your man page. Register support is not implemented in man2hlp now. -- Regards, Andrew V. Samoilov From kai at cmail.ru Mon Oct 21 17:40:01 2002 From: kai at cmail.ru (Andrew V. Samoilov) Date: Mon, 21 Oct 2002 20:40:01 +0300 Subject: preserve_uidgid is described in the manual, but is not implemented Message-ID: <3DB43BF1.50408@cmail.ru> This feature should be usefull for system administrators and even documented in the mc.1, but it is commented out in setup.c and does not implemented in filegui.c and fileopctx.c. Is it worth the trouble to implement this feature? (about 10 lines in the 3 files). -- Regards, Andrew V. Samoilov From proski at gnu.org Mon Oct 21 21:33:19 2002 From: proski at gnu.org (Pavel Roskin) Date: Mon, 21 Oct 2002 17:33:19 -0400 (EDT) Subject: "Safe delete" featured in Interface Hall of Shame In-Reply-To: Message-ID: > As an intermediate solution, I suggest that "Safe delete" stays, but its > only effect will be changing the default button to "no" in the delete > dialogs. Done. The manuals have been updated. -- Regards, Pavel Roskin From proski at gnu.org Mon Oct 21 23:50:42 2002 From: proski at gnu.org (Pavel Roskin) Date: Mon, 21 Oct 2002 19:50:42 -0400 (EDT) Subject: Using poptPrintHelp() In-Reply-To: <3DAE6210.10106@cmail.ru> Message-ID: Hello! > Now it is in stream to use poptSetOtherOptionHelp() in print_mc_usage(), > but this change requires massive po/*.po updates. Applied. The massive changes have beed done. By the way, don't call functions inside fprintf() - it can be a macro, and it can evaluate it's arguments more than once. -- Regards, Pavel Roskin From proski at gnu.org Tue Oct 22 04:38:48 2002 From: proski at gnu.org (Pavel Roskin) Date: Tue, 22 Oct 2002 00:38:48 -0400 (EDT) Subject: Problem with Red Hat S-Lang worked around Message-ID: Hello! It is now possible to link mc against the patched version of S-Lang from Red Hat 8.0 without losing non-ASCII characters in the menu. For some reason, Red Hat S-Lang expects the argument to addch() to be a wide character (wchar_t), so I had to write a replacement to addch(). It is not UTF-8 support, not even a workaround for UTF-8 terminals and locales. It's merely a workaround for a bug in S-Lang, which changes the API, but doesn't change the function name and doesn't even provide a simpler way to use the old API. The current version of mc still only works correctly in non-UTF locales on non-UTF terminals. The only workaround for UTF-8 is to disable 8-bit output (Options->Display Bits->7 bits) and to use "stickchars" (mc -a). I'm not planning adding UTF-8 support. Those who want it should do it themselves. We are now just 9 TODO items away from 4.6.0-pre2. -- Regards, Pavel Roskin From proski at gnu.org Tue Oct 22 05:04:43 2002 From: proski at gnu.org (Pavel Roskin) Date: Tue, 22 Oct 2002 01:04:43 -0400 (EDT) Subject: preserve_uidgid is described in the manual, but is not implemented In-Reply-To: <3DB43BF1.50408@cmail.ru> Message-ID: Hello! > This feature should be usefull for system administrators and even > documented in the mc.1, but it is commented out in setup.c and does not > implemented in filegui.c and fileopctx.c. > > Is it worth the trouble to implement this feature? (about 10 lines in > the 3 files). Indeed, the global preserve_uidgid is unused. There is a field in FileOpContext with this name, but it's always set for root and unset for other users. You can only implement this option for root, and it probably belongs to the file dialogs. Global config-only options are very inconvenient to use, especially an option like preserve_uidgid. The sysadmin does different tasks. Some require preserving the ownership, some don't. Changing /root/.mc/ini every time is ugly. root is not a user. It's a position. Ideally, root should not have user preferences, because different users can be using this account for administration. It's just wrong to say something like "root on this machine prefers to preserve ownership of files when copying them". I'd like to get rid of most hidden options, and preserve_uidgid seems to be a good candidate to be killed now. There is no point in keeping documentation about it - if we ever reimplement this option, it will be a part of a dialog, and will require a different description. -- Regards, Pavel Roskin From kai at cmail.ru Wed Oct 23 09:42:17 2002 From: kai at cmail.ru (Andrew V. Samoilov) Date: Wed, 23 Oct 2002 12:42:17 +0300 Subject: (fwd) ncurses vs slang Message-ID: <20021023124217.A16230@itws8.gca.usb.zp.ua> Hi! These mails from mutt-dev at dbnet.net can be related to ncurses 5.2 turns off keypad under heavy load on xterm. Check if anything can be done about it. See last mail in link and thanks to Andrew W. Nosenko for sending these mails to me. ----- Forwarded message from "Andrew W. Nosenko" ----- X-UIDL: 3d93f11800000003 X-Mozilla-Status: 0001 X-Mozilla-Status2: 10000000 Date: Thu, 26 Sep 2002 19:01:14 +0300 From: "Andrew W. Nosenko" To: sav at bcs.zp.ua Subject: (fwd) ncurses vs slang User-Agent: Mutt/1.5.1i -- Andrew W. Nosenko (awn at bcs.zp.ua) Delivered-To: mutt-dev at ns.gbnet.net Date: Wed, 18 Sep 2002 20:07:24 -0700 From: Burton Samograd To: mutt-dev at gbnet.net Subject: ncurses vs slang Mail-Followup-To: mutt-dev at gbnet.net X-GPG-key: http://kruhftwerk.dyndns.org/kruhft.pubkey.asc X-Operating-System: Linux kruhft.dyndns.org 2.4.19-gentoo-r7 User-Agent: Mutt/1.5.1i Precedence: bulk List-Unsubscribe: Hi all, I was just wondering if there was any advantage to using slang over ncurses. burton Delivered-To: mutt-dev at ns.gbnet.net Date: Thu, 19 Sep 2002 05:14:36 -0400 (EDT) From: "Thomas E. Dickey" Reply-To: To: Burton Samograd cc: Subject: Re: ncurses vs slang In-Reply-To: <20020919030724.GA5495 at kruhft.dyndns.org> Precedence: bulk List-Unsubscribe: On Wed, 18 Sep 2002, Burton Samograd wrote: > Hi all, > > I was just wondering if there was any advantage to using slang over > ncurses. no one's pointed out any (none that count) -- T.E.Dickey http://invisible-island.net ftp://invisible-island.net Delivered-To: mutt-dev at ns.gbnet.net Date: Thu, 19 Sep 2002 19:07:41 +0200 From: Sven Guckes To: mutt-dev at gbnet.net Subject: Re: ncurses vs slang Mail-Followup-To: mutt-dev at gbnet.net In-Reply-To: <20020919030724.GA5495 at kruhft.dyndns.org> User-Agent: Mutt/1.4i X-Envelope-Sender: guckes at math.fu-berlin.de X-Virus-Scanned: by AMaViS 0.3.12pre7-U7 [3681] (NAI-uvscan at math.fu-berlin.de) X-Remote-IP: 160.45.41.21 Precedence: bulk List-Unsubscribe: * Burton Samograd [2002-09-19 03:07]: > I was just wondering if there was any > advantage to using slang over ncurses. "slang" is shorter than "ncurses" - but you have to add "--with-slang" when configuring mutt for compilation. so.... Sven [well, he asked, right?] Delivered-To: mutt-dev at ns.gbnet.net Date: Thu, 19 Sep 2002 13:16:49 -0400 (EDT) From: "Thomas E. Dickey" Reply-To: To: Sven Guckes cc: Subject: Re: ncurses vs slang In-Reply-To: <20020919170741.GR9959 at ritz.math.fu-berlin.de> Precedence: bulk List-Unsubscribe: On Thu, 19 Sep 2002, Sven Guckes wrote: > * Burton Samograd [2002-09-19 03:07]: > > I was just wondering if there was any > > advantage to using slang over ncurses. > > "slang" is shorter than "ncurses" - ncurses hmm (yes - but the library is bigger, last time I looked) > but you have to add "--with-slang" > when configuring mutt for compilation. > so.... > > Sven [well, he asked, right?] > -- T.E.Dickey http://invisible-island.net ftp://invisible-island.net Delivered-To: mutt-dev at ns.gbnet.net Date: Mon, 23 Sep 2002 13:58:50 +0300 From: Jussi Ekholm To: mutt-dev at gbnet.net Subject: Re: ncurses vs slang Mail-Followup-To: mutt-dev at gbnet.net In-Reply-To: <20020919170741.GR9959 at ritz.math.fu-berlin.de> X-Editor: Vim 6.1.174 (CVS) - http://www.vim.org/ X-Operating-System: Linux 2.4.18 X-PGP-Key: http://erppimaa.ihku.org/~ekhowl/pubkey.asc X-URL: http://erppimaa.ihku.org/ User-Agent: Mutt/1.5.1i Precedence: bulk List-Unsubscribe: -----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Sven Guckes wrote: > * Burton Samograd [2002-09-19 03:07]: >> I was just wondering if there was any >> advantage to using slang over ncurses. > > "slang" is shorter than "ncurses" - but you have to add > "--with-slang" when configuring mutt for compilation. so.... You have been getting worse and worse lately, Sven -- are you perhaps getting [too] old or something...? :-) - -- Jussi Ekholm | | <0x1410081E> -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.0.7 (GNU/Linux) iD8DBQE9jvPqAtEARxQQCB4RAv9+AJ9eyQvD4pHXfDfB3AJ8CGekaysyqgCg0Sl4 zDoCe74jxyL7JJh29juZEzM= =P+cP -----END PGP SIGNATURE----- Delivered-To: mutt-dev at ns.gbnet.net Date: Mon, 23 Sep 2002 20:54:40 +0200 (CEST) From: Alain Bench To: Mutt dev ml Subject: Re: ncurses vs slang Mail-Followup-To: Mutt dev ml In-Reply-To: <20020919030724.GA5495 at kruhft.dyndns.org> User-Agent: Mutt/1.4i Precedence: bulk List-Unsubscribe: Hello Burton, On Wednesday, September 18, 2002 at 8:07:24 PM -0700, Burton Samograd wrote: > To: mutt-dev at gbnet.net > Mail-Followup-To: mutt-dev at gbnet.net Better use the list's prefered address: mutt-dev at mutt.org > I was just wondering if there was any advantage to using slang over > ncurses. I haven't precise timings, but I once noticed Slang was generally a little bit faster at redraws, scrolling, and such things. Bye! Alain. Delivered-To: mutt-dev at ns.gbnet.net Date: Mon, 23 Sep 2002 18:37:04 -0400 From: Thomas Dickey To: Mutt dev ml Subject: Re: ncurses vs slang Reply-To: dickey at herndon4.his.com Mail-Followup-To: Thomas Dickey , Mutt dev ml In-Reply-To: <20020923185439.GA3404 at free.fr> User-Agent: Mutt/1.3.27i Precedence: bulk List-Unsubscribe: On Mon, Sep 23, 2002 at 08:54:40PM +0200, Alain Bench wrote: > Hello Burton, > > On Wednesday, September 18, 2002 at 8:07:24 PM -0700, Burton Samograd wrote: > > > To: mutt-dev at gbnet.net > > Mail-Followup-To: mutt-dev at gbnet.net > > Better use the list's prefered address: mutt-dev at mutt.org > > > > I was just wondering if there was any advantage to using slang over > > ncurses. > > I haven't precise timings, but I once noticed Slang was generally > a little bit faster at redraws, scrolling, and such things. most of that was because slang doesn't pay attention to the padding specified in the vt100 entry (the numbers in <>): # Reconstructed via infocmp from file: /usr/local/ncurses/lib/terminfo/v/vt100 vt100|vt100-am|dec vt100 (w/advanced video), am, msgr, xenl, xon, cols#80, it#8, lines#24, vt#3, acsc=``aaffggjjkkllmmnnooppqqrrssttuuvvwwxxyyzz{{||}}~~, bel=^G, blink=\E[5m$<2>, bold=\E[1m$<2>, clear=\E[H\E[J$<50>, cr=^M, csr=\E[%i%p1%d;%p2%dr, cub=\E[%p1%dD, cub1=^H, cud=\E[%p1%dB, cud1=^J, cuf=\E[%p1%dC, cuf1=\E[C$<2>, cup=\E[%i%p1%d;%p2%dH$<5>, cuu=\E[%p1%dA, cuu1=\E[A$<2>, ed=\E[J$<50>, el=\E[K$<3>, el1=\E[1K$<3>, enacs=\E(B\E)0, home=\E[H, ht=^I, hts=\EH, ind=^J, ka1=\EOq, ka3=\EOs, kb2=\EOr, kbs=^H, kc1=\EOp, kc3=\EOn, kcub1=\EOD, kcud1=\EOB, kcuf1=\EOC, kcuu1=\EOA, kent=\EOM, kf0=\EOy, kf1=\EOP, kf10=\EOx, kf2=\EOQ, kf3=\EOR, kf4=\EOS, kf5=\EOt, kf6=\EOu, kf7=\EOv, kf8=\EOl, kf9=\EOw, rc=\E8, rev=\E[7m$<2>, ri=\EM$<5>, rmacs=^O, rmam=\E[?7l, rmkx=\E[?1l\E>, rmso=\E[m$<2>, rmul=\E[m$<2>, rs2=\E>\E[?3l\E[?4l\E[?5l\E[?7h\E[?8h, sc=\E7, sgr=\E[0%?%p1%p6%|%t;1%;%?%p2%t;4%;%?%p1%p3%|%t;7%;%?%p4%t;5%;m%?%p9%t\016%e\017%;$<2>, sgr0=\E[m\017$<2>, smacs=^N, smam=\E[?7h, smkx=\E[?1h\E=, smso=\E[7m$<2>, smul=\E[4m$<2>, tbc=\E[3g, See NCURSES_NO_PADDING in the ncurses manpage... -- Thomas E. Dickey http://invisible-island.net ftp://invisible-island.net ----- End forwarded message ----- From kai at cmail.ru Wed Oct 23 13:31:43 2002 From: kai at cmail.ru (Andrew V. Samoilov) Date: Wed, 23 Oct 2002 16:31:43 +0300 Subject: Backward search in the internal viewer Message-ID: <3DB6A4BF.1080609@cmail.ru> Hello, this feature is documented, but I don't know was this one implemented before 4.1. I implemented it for normal and hex search in 2000, but it was rejected because it had not regular search. I have no ideas how to implement this search for ^[a-b], ^[^c-G\]\] and ^(bif|foo)? cases without much efforts. Any ideas? -- Regards, Andrew V. Samoilov From proski at gnu.org Wed Oct 23 13:52:31 2002 From: proski at gnu.org (Pavel Roskin) Date: Wed, 23 Oct 2002 09:52:31 -0400 (EDT) Subject: Backward search in the internal viewer In-Reply-To: <3DB6A4BF.1080609@cmail.ru> Message-ID: On Wed, 23 Oct 2002, Andrew V. Samoilov wrote: > Hello, > > this feature is documented, but I don't know was this one implemented > before 4.1. I implemented it for normal and hex search in 2000, > but it was rejected because it had not regular search. > > I have no ideas how to implement this search for ^[a-b], ^[^c-G\]\] and > ^(bif|foo)? cases without much efforts. I didn't realize at that time that regex search backwards is hard. Please resend your patch or just apply it if you are certain that it's OK. I remember it also included case sensitive support. That would be really great. New users are not supposed to know if the search in the viewer is case sensitive or not. -- Regards, Pavel Roskin From kai at cmail.ru Wed Oct 23 15:18:14 2002 From: kai at cmail.ru (Andrew V. Samoilov) Date: Wed, 23 Oct 2002 18:18:14 +0300 Subject: Backward search in the internal viewer References: Message-ID: <3DB6BDB6.7080501@cmail.ru> Pavel Roskin wrote: > On Wed, 23 Oct 2002, Andrew V. Samoilov wrote: > > >>Hello, >> >>this feature is documented, but I don't know was this one implemented >>before 4.1. I implemented it for normal and hex search in 2000, >>but it was rejected because it had not regular search. >> >>I have no ideas how to implement this search for ^[a-b], ^[^c-G\]\] and >>^(bif|foo)? cases without much efforts. > > > I didn't realize at that time that regex search backwards is hard. > Please resend your patch or just apply it if you are certain that it's OK. > I remember it also included case sensitive support. That would be really No, it was patch to implement case insensitive context search in "Find file". But I plan to implement case sensitive search support for viewer. > great. New users are not supposed to know if the search in the viewer is > case sensitive or not. -- Regards, Andrew V. Samoilov From proski at gnu.org Wed Oct 23 16:21:22 2002 From: proski at gnu.org (Pavel Roskin) Date: Wed, 23 Oct 2002 12:21:22 -0400 (EDT) Subject: (fwd) ncurses vs slang In-Reply-To: <20021023124217.A16230@itws8.gca.usb.zp.ua> Message-ID: Hello! On Wed, 23 Oct 2002, Andrew V. Samoilov wrote: > These mails from mutt-dev at dbnet.net can be related to > > ncurses 5.2 turns off keypad under heavy load on xterm. Check if > anything can be done about it. > > See last mail in link and thanks to Andrew W. Nosenko for > sending these mails to me. Thanks, but keypad is used for keyboard input, and this thread is about screen output. Besides, last time I checked, the keypad bug was specific to xterm and could not be reproduced on rxvt with the same TERM, so any misinterpretation of terminfo can be probably ruled out. It's just a matter of linking with with static non-optimized ncurses and running keypad() through the debugger. I don't expect it to be very hard, unless we are forced to reimplement some parts of ncurses. -- Regards, Pavel Roskin From kai at cmail.ru Wed Oct 23 16:21:24 2002 From: kai at cmail.ru (Andrew V. Samoilov) Date: Wed, 23 Oct 2002 19:21:24 +0300 Subject: Backward search in the internal viewer References: Message-ID: <3DB6CC84.8010908@cmail.ru> Pavel Roskin wrote: > On Wed, 23 Oct 2002, Andrew V. Samoilov wrote: > > I didn't realize at that time that regex search backwards is hard. > Please resend your patch or just apply it if you are certain that it's OK. I resend this patch, please test it for some days before applying, I sow some strange effects in the hex editor (tab was not working and highlighting was wrong) but cannot reproduce it and I cannot say it is side effect of this patch. Also is it right to change direction by clicking "Backward" check button and entering empty string as search string? -- Regards, Andrew V. Samoilov -------------- next part -------------- An embedded and charset-unspecified text was scrubbed... Name: view.bksearch-2002-10-23.patch URL: From proski at gnu.org Wed Oct 23 16:31:23 2002 From: proski at gnu.org (Pavel Roskin) Date: Wed, 23 Oct 2002 12:31:23 -0400 (EDT) Subject: Problem with Red Hat S-Lang worked around In-Reply-To: Message-ID: Hello! Replying to myself. > It is now possible to link mc against the patched version of S-Lang from > Red Hat 8.0 without losing non-ASCII characters in the menu. For some > reason, Red Hat S-Lang expects the argument to addch() to be a wide > character (wchar_t), so I had to write a replacement to addch(). I was primarily checking the menus (in ru_RU.KOI8-R locale), but the editor, the viewer and non-ASCII files on panels are still broken in C locale (en_US is OK) :-( Re-added to TODO. My preliminary understanding is that conversions to Unicode and back still happens, and they kill non-ASCII characters in the C locale. It may happen that the only way to get rid of those conversions it to add some significant parts of non-UTF code (SLsmg_write_nchars) to mc. Not very nice. Another solution is to stop recognizing Red Hat S-Lang in configure until it's fixed. > We are now just 9 TODO items away from 4.6.0-pre2. 10 again :-( -- Regards, Pavel Roskin From proski at gnu.org Thu Oct 24 05:18:21 2002 From: proski at gnu.org (Pavel Roskin) Date: Thu, 24 Oct 2002 01:18:21 -0400 (EDT) Subject: Links -dump as HTML viewer In-Reply-To: <20021017114235.938AC3E11@xmxpita.excite.com> Message-ID: Hi, David! Sorry, it took a while before I could reply. > We may have to give up and use lynx again. Any idea? Any contact in the > links development team? It's a bit unfair to remove such feature... I have removed links from the viewer associations for HTML files. Please don't consider having it there a feature. mc.ext is by its design just a template for the users. It was implemented long ago, and when it was implemented, it was common for the users to create the associations themselves. The existing code is just not designed to be a comprehensive database of associations. If it was, mc would at least check if the program is in the path, either when the action is invoked by the user or when the local extension file is generated from the global template. As it often happens, the old code was fitted for the new users who expected the associations to work out-of-box. But this just cannot be done reliably. We are simply returning to the original idea of mc.ext. This file is supposed to be modified by users. There is no point in contacting links developers. It's a problem of mc. Rewrite of mc.ext support is planned after version 4.6.0. -- Regards, Pavel Roskin From proski at gnu.org Thu Oct 24 06:05:41 2002 From: proski at gnu.org (Pavel Roskin) Date: Thu, 24 Oct 2002 02:05:41 -0400 (EDT) Subject: links in man2hlp In-Reply-To: <3DB3E025.103@cmail.ru> Message-ID: Hello! > > I tried this to get punctuation marks out the link references > > avoiding extra spaces (ala .BR or .IR). The best idea I got > > was using "\&" as separator because it seems to have no side > > effects for groff in the man page. I read something about > > "\&" preventing kerning between characters but that doesn't > > seem anything to worry about. > > Another possible solution is .\" at the same line in .\"LINK2 Please show the examples. > > I invented a LINK3 entry but if you trust the code it could go > > as LINK2. I was careful with strings but I don't know if > > glib functions must be used anyway. Please, let's try to reduce complexity. Two link formats is too much. If somebody applies your patch, then your code is trusted already. Also please comment every fixme you add. > > Is there anyway to get the date into the help/groff files? > > I (me again :) wanted to complete the header and the footer > > (in .TH at the head of the file). @VERSION@ seems to do the > > job for the version (like @prefix@) in mc.1.in. Why do you want to put the date automatically? What's the point in having any date in the manual if the date is updated automatically? Updating it by hand would be an extra requirement for the maintainers who edit the manual. Am I missing something? Is the date required? > There are yr (years since 1900), mo (month) and dy (day of > month)registers in groff. So, you can add something like > .nr yr \n(yr+1900 > .TH ... "\n(dy-\n(mo-\n(yr" > at the beginning of your man page. > > Register support is not implemented in man2hlp now. It shouldn't matter, since man2hlp happily skips the ".TH" macro. It would have to skip .nr as well, but I don't like this idea. -- Regards, Pavel Roskin From kai at cmail.ru Thu Oct 24 06:56:13 2002 From: kai at cmail.ru (Andrew V. Samoilov) Date: Thu, 24 Oct 2002 09:56:13 +0300 Subject: links in man2hlp References: Message-ID: <3DB7998D.3090005@cmail.ru> Pavel Roskin wrote: > Hello! > > >>>I tried this to get punctuation marks out the link references >>>avoiding extra spaces (ala .BR or .IR). The best idea I got >>>was using "\&" as separator because it seems to have no side >>>effects for groff in the man page. I read something about >>>"\&" preventing kerning between characters but that doesn't >>>seem anything to worry about. >> >>Another possible solution is .\" at the same line in .\"LINK2 > > > Please show the examples. .\"LINK2" Layout\&). .\"Layout" Layout should be highlighted, but ). - no. \& really is best, and no need to provide .\"LINK3". BTW, we need something like README.man2hlp for translators. These people have not to be programmers and have not time to read man2hlp.c. Well, who can explain me is doublequotes required in .I "\-b" ? I removed all of these '"' in the Russian manual, and I can't see difference after groff -mandoc ... Also "Active VFS list", "Free VFSs now", "Menu ediTor edit", "Syntax file edit" from Command menu are not documented in manual, as far as "Charset Conversion". > >>>Is there anyway to get the date into the help/groff files? >>>I (me again :) wanted to complete the header and the footer >>>(in .TH at the head of the file). @VERSION@ seems to do the >>>job for the version (like @prefix@) in mc.1.in. >> > > Why do you want to put the date automatically? What's the point in having > any date in the manual if the date is updated automatically? Updating it > by hand would be an extra requirement for the maintainers who edit the > manual. > > Am I missing something? Is the date required? Sometimes it is so, but manual is not the case. -- Regards, Andrew V. Samoilov From kai at cmail.ru Thu Oct 24 07:56:47 2002 From: kai at cmail.ru (Andrew V. Samoilov) Date: Thu, 24 Oct 2002 10:56:47 +0300 Subject: links in man2hlp References: <3DB7998D.3090005@cmail.ru> Message-ID: <3DB7A7BF.7000007@cmail.ru> Andrew V. Samoilov wrote: > Pavel Roskin wrote: > >> Hello! >> >> >>>> I tried this to get punctuation marks out the link references >>>> avoiding extra spaces (ala .BR or .IR). The best idea I got >>>> was using "\&" as separator because it seems to have no side >>>> effects for groff in the man page. I read something about >>>> "\&" preventing kerning between characters but that doesn't >>>> seem anything to worry about. >>> >>> >>> Another possible solution is .\" at the same line in .\"LINK2 >> >> >> >> Please show the examples. > > > .\"LINK2" > Layout\&). > .\"Layout" > > Layout should be highlighted, but ). - no. > \& really is best, and no need to provide .\"LINK3". BTW, we need > something like README.man2hlp for translators. These people have not to > be programmers and have not time to read man2hlp.c. > Current solution is to move ")." two lines below, but this produce extra space between "Layout" and ")." in the manual. I produced simple patch for man2hlp.c. ChangeLog entry should be Englished ;-) ChangeLog: * man2hlp.c (handle_link): Print part after \& in the linkname behind linkname. Reported by David Martin . -- Regards, Andrew V. Samoilov -------------- next part -------------- An embedded and charset-unspecified text was scrubbed... Name: man2hlp.diff URL: From dmartina at excite.com Thu Oct 24 11:51:00 2002 From: dmartina at excite.com (David Martin) Date: Thu, 24 Oct 2002 07:51:00 -0400 (EDT) Subject: links in man2hlp Message-ID: <20021024115100.EA987109EAE@xmxpita.excite.com> --- On Thu 10/24, Pavel Roskin proski at gnu.org wrote: > Why do you want to put the date automatically? What's the >point in > having > any date in the manual if the date is updated automatically? Updating it > by hand would be an extra requirement for the maintainers who >edit the > manual. > > Am I missing something? Is the date required? I was thinking of the man page. Date is usually in the header or footer of most UNIX pages (I don't know if there's a standard about it). By now, we have in mc.1.in: .TH mc 1 "30 October 1998" As we are making many changes, I think we should update it. We can do it once for all help files, do it automatically or let translators put there whatever they like. I would propose the first choice, but I was checking if second was possible. *8-) David _______________________________________________ Join Excite! - http://www.excite.com The most personalized portal on the Web! From japc at co.sapo.pt Thu Oct 24 11:05:51 2002 From: japc at co.sapo.pt (Jose Celestino) Date: Thu, 24 Oct 2002 12:05:51 +0100 Subject: [BUG] running mc "over" running one Message-ID: <20021024110551.GC1820@co.sapo.pt> It there any known bug running mc over an existing one frozing the virtual terminal? I mean, when I'm running mc on a xterm and I do a CTRL+O and (by mistake) I run another mc get the xterm to froze. -- Jose Celestino SysAdmin::SAPO.pt http://www.sapo.pt ----------------------------------------------------- "Lately, the only thing keeping me from becoming a serial killer is my distaste for manual labor." -- Dilbert From proski at gnu.org Thu Oct 24 12:35:26 2002 From: proski at gnu.org (Pavel Roskin) Date: Thu, 24 Oct 2002 08:35:26 -0400 (EDT) Subject: [BUG] running mc "over" running one In-Reply-To: <20021024110551.GC1820@co.sapo.pt> Message-ID: Hello! > It there any known bug running mc over an existing one frozing the > virtual terminal? As far as I know, it's not a known bug. > I mean, when I'm running mc on a xterm and I do a CTRL+O and (by > mistake) I run another mc get the xterm to froze. I could not reproduce this bug with the CVS version of mc and xterm from XFree86 4.2.0. Please report the bug proprerly. Recommended reading: http://www.chiark.greenend.org.uk/~sgtatham/bugs.html -- Regards, Pavel Roskin From Dmitry.Semyonov at oktet.ru Thu Oct 24 13:05:45 2002 From: Dmitry.Semyonov at oktet.ru (Dmitry Semyonov) Date: Thu, 24 Oct 2002 17:05:45 +0400 (MSD) Subject: File copy related bugs Message-ID: Hello! I've just discovered some more bugs... 1. * 'Options->Configuration->Compute totals' and 'Options->Configuration->Verbose operation' are enabled. * BUG: totals are not computed when copying _single directory_ (if _two ore more dirs_ are copied, then all is ok). 2. * start copying of many files (actually, two dirs with many files) * press 'Skip' button during copying * choose some some answer for 'Incomplete file was retrieved. Keep it?' question. The copying continues after that. * BUG: press left/right arrows to highlight one or another button. You likely to see the same question again, although you have not activated the buttons. 3. Why copy dialog buttons are different for different 'Verbose operation' modes? I mean, 'Abort' button is not presented if 'Verbose operation' is disabled. Well, just found, that it is here actually, but not displayed, untill you press right arrow... Config: $ mc -V bash: cd: GNU Midnight Commander 2002-10-15-16 Virtual File System: tarfs, extfs, cpiofs, ftpfs, fish, undelfs With builtin Editor Using system-installed S-Lang library with terminfo database With subshell support as default With support for background operations With mouse support on xterm and Linux console With support for X11 events With multiple codepages support .: File name too long $ Red Hat Linux release 7.2 (Enigma) Kernel 2.4.19 on an i686 Linux console. ...Bye..Dmitry. From japc at co.sapo.pt Thu Oct 24 13:07:06 2002 From: japc at co.sapo.pt (Jose Celestino) Date: Thu, 24 Oct 2002 14:07:06 +0100 Subject: [BUG] running mc "over" running one In-Reply-To: References: <20021024110551.GC1820@co.sapo.pt> Message-ID: <20021024130706.GI1820@co.sapo.pt> Words by Pavel Roskin [Thu, Oct 24, 2002 at 08:35:26AM -0400]: > Hello! > > > It there any known bug running mc over an existing one frozing the > > virtual terminal? > > As far as I know, it's not a known bug. > > > I mean, when I'm running mc on a xterm and I do a CTRL+O and (by > > mistake) I run another mc get the xterm to froze. > > I could not reproduce this bug with the CVS version of mc and xterm from > XFree86 4.2.0. Please report the bug proprerly. Recommended reading: > Hmm, I can reproduce it every time with xterm 167 (or even konsole from 3.0.3-13) and mc-4.5.55-12 (also with mc-4.6.0-pre1). > http://www.chiark.greenend.org.uk/~sgtatham/bugs.html > > -- > Regards, > Pavel Roskin > -- Jose Celestino SysAdmin::SAPO.pt http://www.sapo.pt ----------------------------------------------------- "Lately, the only thing keeping me from becoming a serial killer is my distaste for manual labor." -- Dilbert From proski at gnu.org Thu Oct 24 15:38:36 2002 From: proski at gnu.org (Pavel Roskin) Date: Thu, 24 Oct 2002 11:38:36 -0400 (EDT) Subject: [BUG] running mc "over" running one In-Reply-To: <20021024130706.GI1820@co.sapo.pt> Message-ID: > Hmm, I can reproduce it every time > > with xterm 167 (or even konsole from 3.0.3-13) and mc-4.5.55-12 (also > with mc-4.6.0-pre1). Please send the output of "mc -V" and the exact steps. That's what I do, trying to follow your instructions: 1) Run xterm 2) Run mc in it 3) Press Ctrl-O 4) Run mc from the command line 5) Press F1. The help shows up. Nothing is frozen. The known problems with running one mc from another are: 1) Only the first mc gets Ctrl-O. That's obvious and won't be fixed. 2) The second mc won't redraw itself when switching to it. The reason is fundamental - there are just two screen buffers in xterm, and mc uses both. The second mc keeps the panels in the buffer that the first mc doesn't restore. The fix is to integrate terminal into mc rather than reuse saving capabilities of the host terminal, but it's hard and not going to happen any time soon. 3) The second mc stops recognizing cursor arrows. This may be a simple bug that needs to be fixed. -- Regards, Pavel Roskin From proski at gnu.org Thu Oct 24 17:04:26 2002 From: proski at gnu.org (Pavel Roskin) Date: Thu, 24 Oct 2002 13:04:26 -0400 (EDT) Subject: links in man2hlp In-Reply-To: <20021024115100.EA987109EAE@xmxpita.excite.com> Message-ID: Hi, David! > As we are making many changes, I think we should update it. We can do it > once for all help files, do it automatically or let translators put > there whatever they like. I would propose the first choice, but I was > checking if second was possible. We can do it before the releases, and only if changes are significant since the last release. Deciding what is "significant" is left to the maintainer and the translators. I have created file maint/RELEASE_PROCEDURE that contains a reminder to the maintainer to check those dates. -- Regards, Pavel Roskin From proski at gnu.org Thu Oct 24 20:13:05 2002 From: proski at gnu.org (Pavel Roskin) Date: Thu, 24 Oct 2002 16:13:05 -0400 (EDT) Subject: File copy related bugs In-Reply-To: Message-ID: Hello! > I've just discovered some more bugs... Thanks! > 1. * 'Options->Configuration->Compute totals' and > 'Options->Configuration->Verbose operation' are enabled. > * BUG: totals are not computed when copying _single directory_ > (if _two ore more dirs_ are copied, then all is ok). Known bug, listed in src/TODO, "before 4.6.1" section: Totals are not calculated when operating on a single directory. > 2. * start copying of many files (actually, two dirs with many files) > * press 'Skip' button during copying > * choose some some answer for 'Incomplete file was retrieved. Keep > it?' question. The copying continues after that. > * BUG: press left/right arrows to highlight one or another button. > You likely to see the same question again, although you have > not activated the buttons. Yes, I can reproduce it. Two hours later... It was hard. My head hurts. All that code really needs to be cleaned up. Debugging it is not fun. I'm applying this patch. ============================== --- ChangeLog +++ ChangeLog @@ -1 +1,7 @@ +2002-10-24 Pavel Roskin + + * filegui.c (check_progress_buttons): Reinitialize ret_value to + avoid old values after incomplete events. + Reported by Dmitry Semyonov + 2002-10-24 Andrew V. Samoilov --- filegui.c +++ filegui.c @@ -170,6 +170,10 @@ check_progress_buttons (FileOpContext *c c = get_event (&event, 0, 0); if (c == EV_NONE) return FILE_CONT; + + /* Reinitialize to avoid old values after incomplete events */ + ui->op_dlg->ret_value = FILE_CONT; + dlg_process_event (ui->op_dlg, c, &event); switch (ui->op_dlg->ret_value) { case FILE_SKIP: ============================== > 3. Why copy dialog buttons are different for different > 'Verbose operation' modes? I mean, 'Abort' button is not presented if > 'Verbose operation' is disabled. Well, just found, that it is here > actually, but not displayed, untill you press right arrow... Yes, very ugly. Too many options, too few testers, and most importantly, too few developers. I'll try to fix it, but you can try it too. -- Regards, Pavel Roskin From proski at gnu.org Thu Oct 24 21:26:38 2002 From: proski at gnu.org (Pavel Roskin) Date: Thu, 24 Oct 2002 17:26:38 -0400 (EDT) Subject: File copy related bugs In-Reply-To: Message-ID: Hello again! > + * filegui.c (check_progress_buttons): Reinitialize ret_value to > + avoid old values after incomplete events. I just realized that it should be "Reinitialize ret_value to avoid old values after events other than selecting a button." Comment fixed. > > 3. Why copy dialog buttons are different for different > > 'Verbose operation' modes? I mean, 'Abort' button is not presented if > > 'Verbose operation' is disabled. Well, just found, that it is here > > actually, but not displayed, untill you press right arrow... That was non-trivial, but I was lucky to find it fast. ========================== --- ChangeLog +++ ChangeLog @@ -2,2 +2,7 @@ + * dlg.c (init_dlg): Redraw screen after selecting the focused + widget, not before. The bug affected file progress dialog with + verbose operation off. + Reported by Dmitry Semyonov + * filegui.c (check_progress_buttons): Reinitialize ret_value to --- dlg.c +++ dlg.c @@ -770,12 +770,13 @@ void init_dlg (Dlg_head *h) /* Initialize the mouse status */ h->mouse_status = MOU_NORMAL; - /* Redraw the screen */ - dlg_redraw (h); - + /* Select the first widget that takes focus */ while (!dlg_focus (h) && h->current) h->current = h->current->next; + /* Redraw the screen */ + dlg_redraw (h); + h->ret_value = 0; h->running = 1; } ========================== -- Regards, Pavel Roskin From proski at gnu.org Thu Oct 24 22:27:06 2002 From: proski at gnu.org (Pavel Roskin) Date: Thu, 24 Oct 2002 18:27:06 -0400 (EDT) Subject: links in man2hlp In-Reply-To: <3DB7A7BF.7000007@cmail.ru> Message-ID: > I produced simple patch for man2hlp.c. ChangeLog entry should be > Englished ;-) Applied. The English manual has been converted. -- Regards, Pavel Roskin From padraig.brady at corvil.com Fri Oct 25 16:16:52 2002 From: padraig.brady at corvil.com (Padraig Brady) Date: Fri, 25 Oct 2002 17:16:52 +0100 Subject: ctime != Change time Message-ID: <3DB96E74.8040202@corvil.com> The sort dialog allows you to sort on: Modification, Access or Change time. bzzt! that should be: Modification, Access or Creation time. P?draig. From padraig.brady at corvil.com Fri Oct 25 16:28:00 2002 From: padraig.brady at corvil.com (Padraig Brady) Date: Fri, 25 Oct 2002 17:28:00 +0100 Subject: new patchfs Message-ID: <3DB97110.1090605@corvil.com> Personally I think the attached is much better than the origonal (details in comments at top) Also it would be nice to add: regex/\.(patch|diff)\.*(gz|bz2)*$ Open=%cd %p#patchfs in the default bindings file (just above ### Archives ###) Also s/lynx/links/ seems appropriate in the default bindings file. cheers, P?draig. -------------- next part -------------- An embedded and charset-unspecified text was scrubbed... Name: patchfs URL: From proski at gnu.org Fri Oct 25 17:35:43 2002 From: proski at gnu.org (Pavel Roskin) Date: Fri, 25 Oct 2002 13:35:43 -0400 (EDT) Subject: ctime != Change time In-Reply-To: <3DB96E74.8040202@corvil.com> Message-ID: Hello! > The sort dialog allows you to sort on: > Modification, Access or Change time. > bzzt! that should be: > Modification, Access or Creation time. . From "info libc" (glibc-2.2.93): `time_t st_ctime' This is the time of the last modification to the attributes of the file. *Note File Times::. If you have references supporting your point of view, please show them. -- Regards, Pavel Roskin From proski at gnu.org Fri Oct 25 17:46:38 2002 From: proski at gnu.org (Pavel Roskin) Date: Fri, 25 Oct 2002 13:46:38 -0400 (EDT) Subject: new patchfs In-Reply-To: <3DB97110.1090605@corvil.com> Message-ID: Hello! On Fri, 25 Oct 2002, Padraig Brady wrote: > Personally I think the attached is much better than the origonal > (details in comments at top) Please send a patch against the CVS version, which can be found here: http://cvs.gnome.org/lxr/source/mc/vfs/extfs/patchfs?raw=1 It seems from your comment that you took the version from 4.5.51 as the base. Version 4.5.51 is years old. Replacing files is very unsafe - it's easy to remove recent changes. > Also it would be nice to add: > > regex/\.(patch|diff)\.*(gz|bz2)*$ > Open=%cd %p#patchfs > > in the default bindings file (just above ### Archives ###) It will be done when either patchfs shows correct file sizes or when the editor stops trusting stat(). patchfs is not ready for users yet. > Also s/lynx/links/ seems appropriate in the default bindings file. You probably didn't see the latest and greatest links that doesn't support the "-dump" option. Check recent discussion in the archives: http://mail.gnome.org/archives/mc-devel/2002-October/msg00047.html -- Regards, Pavel Roskin From kai at cmail.ru Fri Oct 25 18:57:12 2002 From: kai at cmail.ru (Andrew V. Samoilov) Date: Fri, 25 Oct 2002 21:57:12 +0300 Subject: Viewer warns about absent catdoc&word2x _before_ exit for zzz.doc with zzz.doc inside Message-ID: <3DB99408.7040503@cmail.ru> Hello! Andrew W. Nosenko reported strange bug with files named like zzz.doc with only zzz.doc line inside. There is Linux box without catdoc & word2x: $ echo zzz.doc > zzz.doc F3 zzz.doc zzz.doc is showed ok in the viewer. F10, F3, q, etc -- and warning window with message about absent catdoc and word2x appears. mc 4.6.0pre1b. -- Regards, Andrew V. Samoilov. From proski at gnu.org Fri Oct 25 19:23:52 2002 From: proski at gnu.org (Pavel Roskin) Date: Fri, 25 Oct 2002 15:23:52 -0400 (EDT) Subject: Viewer warns about absent catdoc&word2x _before_ exit for zzz.doc with zzz.doc inside In-Reply-To: <3DB99408.7040503@cmail.ru> Message-ID: Hello! > Andrew W. Nosenko reported strange bug with files named > like zzz.doc with only zzz.doc line inside. There is Linux box without > catdoc & word2x: > $ echo zzz.doc > zzz.doc > F3 zzz.doc > zzz.doc is showed ok in the viewer. > F10, F3, q, etc -- and warning window with message about absent catdoc > and word2x appears. I'm not surprized. check_error_pipe() was broken for a long time. Now it's fixed, so that the messages have become visible, and we can see how wrong they are :-) check_error_pipe() is called very late in the viewer - it's called in free_file(), when the viewer exits. We need something like this: ========================= --- view.c +++ view.c @@ -461,6 +461,7 @@ init_growing_view (WView * view, char *n err_msg = (" Empty output from child filter "); return set_view_init_error (view, err_msg); } + check_error_pipe (); } else { view->stdfile = NULL; if ((view->file = mc_open (filename, O_RDONLY)) == -1) ========================= I tried this, and there is another empty warning on exit from the viewer. I don't have time to finish it today (although I often do something after saying that I cannot :-)), and I'll be away on weekend. Incidentally, I think that the *_error_pipe mechanism will be useful in vfs/extfs.c, which also uses popen(). It's better than to see rpm or zip warnings all over the screen. but it means that we should do it right and make empty warnings impossible. -- Regards, Pavel Roskin From makovick at kmlinux.fjfi.cvut.cz Fri Oct 25 20:47:24 2002 From: makovick at kmlinux.fjfi.cvut.cz (Jindrich Makovicka) Date: Fri, 25 Oct 2002 22:47:24 +0200 Subject: [PATCH] ftpfs.c closing the data socket on abort Message-ID: <3DB9ADDC.5070800@kmlinux.fjfi.cvut.cz> Hello, this patch makes MC work with newer versions of ProFTPD (>= 1.2.6) as these versions expect the client to close the data socket. -- Jindrich Makovicka -------------- next part -------------- An embedded and charset-unspecified text was scrubbed... Name: ftpfs.diff URL: From proski at gnu.org Fri Oct 25 21:55:33 2002 From: proski at gnu.org (Pavel Roskin) Date: Fri, 25 Oct 2002 17:55:33 -0400 (EDT) Subject: [PATCH] ftpfs.c closing the data socket on abort In-Reply-To: <3DB9ADDC.5070800@kmlinux.fjfi.cvut.cz> Message-ID: Hello! > this patch makes MC work with newer versions of ProFTPD (>= 1.2.6) as > these versions expect the client to close the data socket Thank you for finding this problem! I've applied your patch because it's obviously right and doesn't break anything for me. As you probably know, testing is more than half of the work when it comes to bugs like this one. I would appreciate if you wrote if there are any public FTP servers that exhibit this problem and how to see if the problem exists (i.e. what goes wrong). It's also helpful to know in case if somebody reports this problem. -- Regards, Pavel Roskin From proski at gnu.org Fri Oct 25 23:13:25 2002 From: proski at gnu.org (Pavel Roskin) Date: Fri, 25 Oct 2002 19:13:25 -0400 (EDT) Subject: [BUG] running mc "over" running one In-Reply-To: <20021024160646.GM1820@co.sapo.pt> Message-ID: Hello, Jose! Let's return the discussion to the list. I've seen some complaints from users reading the archives that the important pieces of the discussion are missing when the discussion becomes private. > > 1) Run xterm 2) Run mc in it 3) Press Ctrl-O 4) Run mc from the > > command line 5) Press F1. The help shows up. Nothing is frozen. > > Exactly the same I do except that I when I get to 5) it is frozen, the > xterm is black. I think it may be a subshell problem, and there are good chances that it's fixed in CVS. At least is was fixed for Cygwin. I have no idea why it happens on your system. Please try the latest snapshot from http://www.ibiblio.org/pub/Linux/utils/file/managers/mc/snapshots/ > 13963 rt_sigprocmask(SIG_BLOCK, [CHLD], [], 8) = 0 13963 > rt_sigsuspend([] There have been significant changes near this point in subshell.c > lrwx------ 1 japc japc 64 Oct 24 17:03 6 -> /dev/pts/12 > lrwx------ 1 japc japc 64 Oct 24 17:03 7 -> /dev/ptmx This means that the second mc has already created the pseudoterminal for its subshell. If it hangs, it happens quite late in the initialization process, most likely in the subshell code. -- Regards, Pavel Roskin From makovick at kmlinux.fjfi.cvut.cz Sat Oct 26 09:18:01 2002 From: makovick at kmlinux.fjfi.cvut.cz (Jindrich Makovicka) Date: Sat, 26 Oct 2002 11:18:01 +0200 Subject: [PATCH] ftpfs socket timeout Message-ID: <3DBA5DC9.3030903@kmlinux.fjfi.cvut.cz> Hello, the following patch imposes send and receive timeouts on the command and data socket. I also changed the vfs_s_select_on_two function so that it can wait for only one descriptor if the second parameter is < 0. Actually it's unclear to me why there was a select on both data socket and stdin performed in both ftpfs and FISH. When I changed it to wait only for the socket, it's now possible to abort the stalled downloads. Previously, mc tried to read from the socket even after pressing a key, which caused a lockup. -- Jindrich Makovicka -------------- next part -------------- An embedded and charset-unspecified text was scrubbed... Name: stall2.diff URL: From pavel at ucw.cz Fri Oct 25 08:23:43 2002 From: pavel at ucw.cz (Pavel Machek) Date: Fri, 25 Oct 2002 10:23:43 +0200 Subject: [BUG] mc edit Floating point exception 4.6.0pre1 In-Reply-To: References: <20021015212142.4438e651.ury@bofh.homeunix.org> Message-ID: <20021025082341.GA24787@elf.ucw.cz> Hi! > > and one more... > > i have trouble in use ftp capabilities under user: > > when i connect to ftp server - i see notning on panel, and in tips > > string i see 'ftpfs: failed; nowhere to fallback to' > > while under r00t - all ok. > > this my mistake or bug? > > There are too little details to reproduce this problem. I remember > similar bugreports in the past. In any case it cannot be your mistake. > I also remember seeing this message myself if the ftp server disconnects > on timeout and I press Ctrl-R to reload the listing. > > The message is so strange because ftpfs tries to use UNIX options for "ls" > and then falls back to more portable options. If that fails as well, this > message is printed. In your case it makes no sense because you don't know > and don't have to know what ftpfs tries to do before it fails. > > It's a reminder for the contributors - try to see the whole picture, not > just the problem your patch tries to solve. I recommend that whoever > submits a patch that changes user-visible strings also provides the > translation to his/her native language. "nowhere to fallback to" may look > very solid when written in English, but try writing this in your language > and think again if your friends will understand what it is about when they > see this message on the screen. It was me, and long time ago. I do not know how to make the message easier -- it should just never ever happen for them. I guess in my native language it would be "nemam co dalsiho zkusit" (actually "nem\'am co dal\vs\'iho zkusit). Pavel -- Worst form of spam? Adding advertisment signatures ala sourceforge.net. What goes next? Inserting advertisment *into* email? From chac02 at student.bth.se Sat Oct 26 08:39:33 2002 From: chac02 at student.bth.se (Christian tricky Andersson) Date: Sat, 26 Oct 2002 10:39:33 +0200 Subject: bug. Message-ID: <20021026103933.278b2157.chac02@student.bth.se> The ftp func. in mc is not working as its supposed to. but you prob. knew this already... It could be like pftp then mc would be perfect. From email at giorgiomantovani.com Sat Oct 26 09:40:05 2002 From: email at giorgiomantovani.com (giorgio_m) Date: Sat, 26 Oct 2002 11:40:05 +0200 Subject: bug mc 4.5.55 Message-ID: <3DBA62F5.5000700@giorgiomantovani.com> mc 4.5.55 (included in mandrake 9) the problem is that when a file name contains a blank space, for example I have this file "photo 1.bmp", I can do nothing with it, I can't open it using F2- at -program another problem, I've got directories names containing spaces, well, if I press F2-make a tar release of the current directory, as always, mc does nothing I can't use mc if these problems aren't fixed, please do something!! thanks mc is great From padraig at linux.ie Sat Oct 26 15:03:19 2002 From: padraig at linux.ie (Padraig Brady) Date: Sat, 26 Oct 2002 16:03:19 +0100 Subject: [PATCH] new patchfs Message-ID: <3DBAAEB7.5000201@linux.ie> On Fri, 25 Oct 2002, Pavel Roskin wrote: > > On Fri, 25 Oct 2002, Padraig Brady wrote: > > > > Personally I think the attached is much better than the origonal > > (details in comments at top) > > Please send a patch against the CVS version, which can be found here: > http://cvs.gnome.org/lxr/source/mc/vfs/extfs/patchfs?raw=1 Not responding? Instead I did: cvs -d :pserver:anonymous at anoncvs.gnome.org:/cvs/gnome co mc/vfs/extfs/patchfs > It seems from your comment that you took the version from 4.5.51 as > the base. Version 4.5.51 is years old. Replacing files is very > unsafe - it's easy to remove recent changes. I checked that the replacement was OK. Anyway the patch is attached. thanks, P?draig. -------------- next part -------------- An embedded and charset-unspecified text was scrubbed... Name: patchfs.diff URL: From padraig at linux.ie Sat Oct 26 17:49:34 2002 From: padraig at linux.ie (Padraig Brady) Date: Sat, 26 Oct 2002 18:49:34 +0100 Subject: better temp file names Message-ID: <3DBAD5AE.6070102@linux.ie> Hi, I just noticed that links (0.96) will not open a html file correctly if it's viewed from within a tar.gz file, but it's fine when viewed from the filesystem. This is because mc copies the file to something like /tmp/m273hiu37 which links will not treat like a html file by default as it doesn't have a html extension. I know links shouldn't behave like this or should at least have a force-html option but to minimise problems like this wouldn't it be better to use the extension of the origonal filename for the tmp file, in this case /tmp/m273hiu37.html P?draig. From proski at gnu.org Sun Oct 27 01:12:13 2002 From: proski at gnu.org (Pavel Roskin) Date: Sat, 26 Oct 2002 21:12:13 -0400 Subject: better temp file names In-Reply-To: <3DBAD5AE.6070102@linux.ie> References: <3DBAD5AE.6070102@linux.ie> Message-ID: <1035681133.3dbb3d6d64b67@webmail.spamcop.net> An embedded and charset-unspecified text was scrubbed... Name: not available URL: From proski at gnu.org Sun Oct 27 01:22:45 2002 From: proski at gnu.org (Pavel Roskin) Date: Sat, 26 Oct 2002 21:22:45 -0400 Subject: bug. In-Reply-To: <20021026103933.278b2157.chac02@student.bth.se> References: <20021026103933.278b2157.chac02@student.bth.se> Message-ID: <1035681765.3dbb3fe527970@webmail.spamcop.net> An embedded and charset-unspecified text was scrubbed... Name: not available URL: From proski at gnu.org Sun Oct 27 01:30:30 2002 From: proski at gnu.org (Pavel Roskin) Date: Sat, 26 Oct 2002 21:30:30 -0400 Subject: bug mc 4.5.55 In-Reply-To: <3DBA62F5.5000700@giorgiomantovani.com> References: <3DBA62F5.5000700@giorgiomantovani.com> Message-ID: <1035682230.3dbb41b6ec2ef@webmail.spamcop.net> An embedded and charset-unspecified text was scrubbed... Name: not available URL: From proski at gnu.org Sun Oct 27 01:36:45 2002 From: proski at gnu.org (Pavel Roskin) Date: Sat, 26 Oct 2002 21:36:45 -0400 Subject: [PATCH] new patchfs In-Reply-To: <3DBAAEB7.5000201@linux.ie> References: <3DBAAEB7.5000201@linux.ie> Message-ID: <1035682605.3dbb432d7c2db@webmail.spamcop.net> An embedded and charset-unspecified text was scrubbed... Name: not available URL: From proski at gnu.org Mon Oct 28 08:04:38 2002 From: proski at gnu.org (Pavel Roskin) Date: Mon, 28 Oct 2002 03:04:38 -0500 (EST) Subject: [BUG] mc edit Floating point exception 4.6.0pre1 In-Reply-To: <20021025082341.GA24787@elf.ucw.cz> Message-ID: Hello! > It was me, and long time ago. I do not know how to make the message > easier -- it should just never ever happen for them. I guess in my > native language it would be "nemam co dalsiho zkusit" (actually > "nem\'am co dal\vs\'iho zkusit). It was simply an example to show future developers a typical error coming from not thinking of the interface from the user point of view. Your explanation shows that it wasn't a good example. Sorry for that. That's not to say that designing the interface with the user in mind is not important. -- Regards, Pavel Roskin From dmartina at excite.com Mon Oct 28 10:46:07 2002 From: dmartina at excite.com (David Martin) Date: Mon, 28 Oct 2002 05:46:07 -0500 (EST) Subject: Docs updated & some GPL issues Message-ID: <20021028104607.740EABF78@xmxpita.excite.com> ------------------------- About docs - Added some in Spanish help links and clone latest changes from English file. More updating, as usual. - Spanish xnc.hlp: Adding GNU everywhere and make it more readable. - es.po and es_ES.po updated - Some extra vertical spacing was left in English mc.1.in - On the other hand, we can't get rid of the .TP at the head of the options if we want proper indenting of first option. GPL -------------------------- - There some minor differences between our COPYING file and the actual GPL. - Help files reproduce these differences and show an outdated address. I made a separate patch to apply it to the different localized help files. - There's a funny ^L shown in the help text GPL in xnc.hlp. Misc ------------------------- - How will the maintainer know how to localize the date in the man pages before the release? - With mc -w you get: -w: unknown option Could we have something like: ? mc: unknown option '-w' Which looks more like in: dd: unrecognized option `-w' Try `dd --help' for more information. or in: cp: invalid option -- w Try `cp --help' for more information. And these programs don't show help with errors. With mc it may be difficult to recognize the real error: This is what I get under Mandrake: ... Please send any bug reports (including the output of `mc -V') to mc-devel at gnome.org -w: unknown option -bash: cd: GNU Midnight Commander 2002-10-25-05 .: No such file or directory ------------------------ Telegraphic enough? *8-) David _______________________________________________ Join Excite! - http://www.excite.com The most personalized portal on the Web! -------------- next part -------------- A non-text attachment was scrubbed... Name: pending14.diff.gz Type: application/x-gzip-compressed Size: 13410 bytes Desc: not available URL: -------------- next part -------------- An embedded and charset-unspecified text was scrubbed... Name: xnc.hlp.diff URL: From dmartina at excite.com Mon Oct 28 10:50:08 2002 From: dmartina at excite.com (David Martin) Date: Mon, 28 Oct 2002 05:50:08 -0500 (EST) Subject: Text formatting with Groff Message-ID: <20021028105008.11050BF58@xmxpita.excite.com> I have an unnofficial Spanish version of GPL which could be included in the help file as others do. I was trying the script below to reformat it. Any advice? Should I make it with Spanish hyphenation? Is it worth? ------------ cat - gples.txt gples62.txt .nh .ll 62 EOF ------------ .nh = no hyphenate .ll = line length *8-) David _______________________________________________ Join Excite! - http://www.excite.com The most personalized portal on the Web! From japc at co.sapo.pt Tue Oct 29 11:06:00 2002 From: japc at co.sapo.pt (Jose Celestino) Date: Tue, 29 Oct 2002 11:06:00 +0000 Subject: [BUG] running mc "over" running one In-Reply-To: References: <20021024160646.GM1820@co.sapo.pt> Message-ID: <20021029110600.GA1096@co.sapo.pt> Words by Pavel Roskin [Fri, Oct 25, 2002 at 07:13:25PM -0400]: > Hello, Jose! > > Let's return the discussion to the list. I've seen some complaints from > users reading the archives that the important pieces of the discussion are > missing when the discussion becomes private. > Fair enough :) > > > 1) Run xterm 2) Run mc in it 3) Press Ctrl-O 4) Run mc from the > > > command line 5) Press F1. The help shows up. Nothing is frozen. > > > > Exactly the same I do except that I when I get to 5) it is frozen, the > > xterm is black. > > I think it may be a subshell problem, and there are good chances that it's > fixed in CVS. At least is was fixed for Cygwin. I have no idea why it > happens on your system. > It is indeed a subshell problem, see ahead. > Please try the latest snapshot from > http://www.ibiblio.org/pub/Linux/utils/file/managers/mc/snapshots/ > All debugging, and everything referenced in this mail is related to this version. > > 13963 rt_sigprocmask(SIG_BLOCK, [CHLD], [], 8) = 0 13963 > > rt_sigsuspend([] > > There have been significant changes near this point in subshell.c > You mean at synchronize? > > lrwx------ 1 japc japc 64 Oct 24 17:03 6 -> /dev/pts/12 > > lrwx------ 1 japc japc 64 Oct 24 17:03 7 -> /dev/ptmx > > This means that the second mc has already created the pseudoterminal for > its subshell. If it hangs, it happens quite late in the initialization > process, most likely in the subshell code. > Exactly. I've done some further debugging and find it to stop after failing at FD_ISSET's in feed_subshell at subshell.c (967 and following -- only the first call to feed_subshell get FD_ISSET ok). It then just returns to init_subshell and gets stuck there. One thing I forgot to include in the previous mail was the state for both mc shell's: [japc at morgoth:/tmp/mc-2002-10-25-22.ALTERED/src]$ ps auxw | grep 'bash -rcfile' japc 7796 0.0 1.1 4632 1416 pts/5 S 03:11 0:00 bash -rcfile .bashrc japc 7802 0.0 0.9 4524 1208 pts/6 T 03:11 0:00 bash -rcfile .bashrc As you can see the second one stays stopped. That is most certainly related with the g_snprintf (precmd, sizeof (precmd), "precmd() {pwd>&%d;kill -STOP $$ }\n", at subshell.c which later gets writen to the pty: write (subshell_pty, precmd, strlen (precmd)); So it is not really the second mc that is stopping/blocking but rather the shell it is trying to execl. Of course that if I do a #undefine HAVE_SUBSHELL_SUPPORT the problems go away. Hope this helps. -- Jose Celestino SysAdmin::SAPO.pt http://www.sapo.pt ----------------------------------------------------- "Lately, the only thing keeping me from becoming a serial killer is my distaste for manual labor." -- Dilbert From proski at gnu.org Tue Oct 29 22:58:31 2002 From: proski at gnu.org (Pavel Roskin) Date: Tue, 29 Oct 2002 17:58:31 -0500 (EST) Subject: Docs updated & some GPL issues In-Reply-To: <20021028104607.740EABF78@xmxpita.excite.com> Message-ID: Hi, David! I've applied your patches. Thank you! Do you have write access to CVS now? > - On the other hand, we can't get rid of the .TP at the head of the > options if we want proper indenting of first option. Thanks! My error. > - There some minor differences between our COPYING file and the actual > GPL. I've removed it from CVS, so that automake will install it. > - Help files reproduce these differences and show an outdated address. I > made a separate patch to apply it to the different localized help files. Maybe man2hlp should insert the COPYING file automatically? > - There's a funny ^L shown in the help text GPL in xnc.hlp. It's present in the text on GPL. Another reason not to trust humans (to create xnc.hlp) :-) > - How will the maintainer know how to localize the date in the man pages > before the release? I have fixed maint/RELEASE_PROCEDURE to make it clear that only the English manual needs to be updated. > - With mc -w you get: > -w: unknown option > > Could we have something like: ? > mc: unknown option '-w' I wish we could, but "unknown option" in poptStrerror() has no "%s" and some languages may require the option name to be in the middle or even in the beginning. You cannot glue '-w' at the end without ":" and expect it to sound right in every language. How about this? (Should be easy to do, see handle_args() in main.c) mc: unknown option: '-w' > This is what I get under Mandrake: > -bash: cd: GNU Midnight Commander 2002-10-25-05 > .: No such file or directory It's that ugly -P option. Too bad many people will complain if I remove it. It should write the last directory to a temprorary file, and the shell wrapper should be able to deal with the case when such file is missing. The original idea that mc will write the last directory always was wrong - it makes the code ugly (every time mc exists, it should take care of that stupid script, no matter if mc just prints help), prevents using stdout and creates a security risk if some subprocesses write something bad to stdout. Fixing -P is in the TODO list. -- Regards, Pavel Roskin From proski at gnu.org Tue Oct 29 23:04:07 2002 From: proski at gnu.org (Pavel Roskin) Date: Tue, 29 Oct 2002 18:04:07 -0500 (EST) Subject: Text formatting with Groff In-Reply-To: <20021028105008.11050BF58@xmxpita.excite.com> Message-ID: Hi, David! > I have an unnofficial Spanish version of GPL which could be > included in the help file as others do. I was trying the > script below to reformat it. Any advice? > Should I make it with Spanish hyphenation? > Is it worth? Sorry, I don't understand what you are trying to do. Are you trying to put it to the manual? That would be wrong. If you just want to reformat it, set "Word wrap line length" to 58 in the mc editor and use Alt-P. Another solution is to run "pico -r58" and use Ctrl-J in it - it moves the cursor down, so you won't have to use cursor arrows. -- Regards, Pavel Roskin From proski at gnu.org Wed Oct 30 01:13:39 2002 From: proski at gnu.org (Pavel Roskin) Date: Tue, 29 Oct 2002 20:13:39 -0500 (EST) Subject: Change in the "Listing mode" documentation Message-ID: Hello! I was trying to make an automated check for the documentation when I found that we have a line in mc.1 that nroff cannot break. That's the description of the long format in the "Listing Mode" section. Also the descriptions of the keywords all included highlighted commas, making it unclear whether commas are required. I have changed the documentation so that it uses spaces, not commas, in the definitions for the listing modes. The default modes in the source have been changed as well. The keyword "mark" has been moved up, closer to the "type" keyword, since it's specific to the file rather than a purely decorative separator. Help in updating the translations will be appreciated. -- Regards, Pavel Roskin From oskar at osk.mine.nu Wed Oct 30 06:09:58 2002 From: oskar at osk.mine.nu (Oskar Liljeblad) Date: Wed, 30 Oct 2002 07:09:58 +0100 Subject: -P fix In-Reply-To: References: <20020905213100.GA17542@oskar> Message-ID: <20021030060958.GA832@oskar> On Wednesday, September 11, 2002 at 03:26, Pavel Roskin wrote: Hi! > > My solution is to add another option -p that will allow you to > > specify a file that the last working directory is written to. > > That's fine. 1. Remove old -P, rename -p to -P. > > A new function print_last_working_directory was created since > > this functionality is needed in two places in main.c. > > > > I also attempted to update the manual page. > > But you didn't update the ouptut of "mc --help". I know, it's wrong that > mc describes options in two different places. It's a result of too many > quick hacks in the sources. But it must stop. 2. Add help for -P everwhere. > You see, you made an error because the existing code is duplicated. We > cannot just ignore this error and go ahead. Otherwise somebody else will > stumble on the same place. Of course not. I agree. > > Here's a new mc.sh: > > Maybe an alias for bash and zsh, since we are breaking everything anyway. > I don't like exposing this function in the environment, since it > discourages developers from improving it (especially increasing it in > size). 3. Make mc.sh a wrapper script instead? Then you could use alias mc=/usr/lib/mc/bin/mc.sh or whatever. > > mc () > > { > > TMPFILE="`mktemp`" > > /usr/bin/mc -p "$TMPFILE" "$@" > > We have a save temporary directory now, so using mktemp is probably not > needed - mc itself can create files safely. If you want MC to create the temporary file, won't you end up with exactly the same issues as before? Or do you intend to make use of a file whose name is known from the beginning? (E.g. $HOME/.mc/mcpwd.$$ where $$ is the pid of the parent processes - the shell.) Regards, Oskar Liljeblad (oskar at osk.mine.nu) From dmartina at excite.com Wed Oct 30 10:55:16 2002 From: dmartina at excite.com (David Martin) Date: Wed, 30 Oct 2002 05:55:16 -0500 (EST) Subject: Text formatting with Groff Message-ID: <20021030105516.3FD64B708@xmxpita.excite.com> I had your same idea about generating xnc.hlp from the GPL original text (automatically). I think that some help (in xnc.hlp, doesn't get to mc.1) translations include both the English and local versions (ru?). *8-) David --- On Tue 10/29, Pavel Roskin wrote: From: Pavel Roskin [mailto: proski at gnu.org] To: dmartina at excite.com Cc: mc-devel at gnome.org Date: Tue, 29 Oct 2002 18:04:07 -0500 (EST) Subject: Re: Text formatting with Groff > Hi, David! > > > I have an unnofficial Spanish version of GPL which could be > > included in the help file as others do. I was trying the > > script below to reformat it. Any advice? > > Should I make it with Spanish hyphenation? > > Is it worth? > > Sorry, I don't understand what you are trying to do. Are you trying to > put it to the manual? That would be wrong. _______________________________________________ Join Excite! - http://www.excite.com The most personalized portal on the Web! From proski at gnu.org Wed Oct 30 21:03:52 2002 From: proski at gnu.org (Pavel Roskin) Date: Wed, 30 Oct 2002 16:03:52 -0500 (EST) Subject: Spanish translations & Mandrake 9.0 report In-Reply-To: <20021014094724.477D23E13@xmxpita.excite.com> Message-ID: Hi, David! > I'm getting a warning about duplicated definitions for something about > the mouse. It's fixed now. I hoped to add wheel support with gpm as well, but I'm too busy and I want to release 4.6.0 without delays. So I just added a few ifdefs for now. The fixed snapshot is being uploaded now. I hope it will compile on Mandrake 9.0 without too many warnings. -- Regards, Pavel Roskin From proski at gnu.org Thu Oct 31 00:23:38 2002 From: proski at gnu.org (Pavel Roskin) Date: Wed, 30 Oct 2002 19:23:38 -0500 (EST) Subject: Backward search in the internal viewer In-Reply-To: <3DB6CC84.8010908@cmail.ru> Message-ID: Hello! I have finally found time to test your patch, but it didn't apply cleanly. I could apply it by hand, but I'm not sure if I did it right. I don't like introducing another preprocessor symbol NO_BK_SEARCH, especially because symbols with the negative meaning (NO_something, DISABLE_this) force the developers to apply an additional logical operation in their heads. I decided to "prepare ground" to make the final patch easier to read. I reformatted view.c and changed charset.h so that it now provides dummy macros to be used when the charset support is disabled. I'm attaching your patch against the new view.c, without and single preprocessor directive. Please check that convert_from_input() is called right, it could have been duplicated when I was applying your patch. I still don't feel I can apply this patch without having more time to review the patch and running it through gdb. But I'll do it after you check the simplified version. -- Regards, Pavel Roskin -------------- next part -------------- A non-text attachment was scrubbed... Name: view-back-search.diff.gz Type: application/x-gzip Size: 1824 bytes Desc: URL: From proski at gnu.org Thu Oct 31 04:49:35 2002 From: proski at gnu.org (Pavel Roskin) Date: Wed, 30 Oct 2002 23:49:35 -0500 (EST) Subject: -P fix In-Reply-To: <20021030060958.GA832@oskar> Message-ID: Hello! > > > My solution is to add another option -p that will allow you to > > > specify a file that the last working directory is written to. > > > > That's fine. > > 1. Remove old -P, rename -p to -P. Is there any reason to rename the short option rather than just require the argument? Should the long option (--printwd) be renamed as well? > 2. Add help for -P everwhere. It's done properly now. > 3. Make mc.sh a wrapper script instead? > Then you could use alias mc=/usr/lib/mc/bin/mc.sh or whatever. I agree. That's what I meant. Perhaps I was unclear. > > We have a save temporary directory now, so using mktemp is probably > > not needed - mc itself can create files safely. > > If you want MC to create the temporary file, won't you end up with > exactly the same issues as before? Or do you intend to make use of a > file whose name is known from the beginning? (E.g. $HOME/.mc/mcpwd.$$ > where $$ is the pid of the parent processes - the shell.) You are right. The filename should be known from the beginning. That doesn't meant the the directory should exist from the beginning. If the parent shell uses a name in the mc temporary directory (/tmp/mc-$USER), the directory will be created by mc if it doesn't exist. I'll try to implement the new scheme step by step. I have already changed the -P option to take an argument. The scripts will need to be rewritten, Then the manual will be changed. -- Regards, Pavel Roskin From proski at gnu.org Thu Oct 31 06:31:11 2002 From: proski at gnu.org (Pavel Roskin) Date: Thu, 31 Oct 2002 01:31:11 -0500 (EST) Subject: Need help with csh syntax Message-ID: Hello! I have converted mc.sh to the new syntax, when the -P option takes an argument - the file where the last working directory is placed by mc. An important fact is that now mc is not guaranteed to create the output file. It is only created when the panels have been initialized. mc.csh needs to be updated as well. I have done simple changes to that file. However, I'm not experienced in csh enough to do it right. In particular, the case when the directory contains spaces or some special symbols (e.g. semicolon) needs to be analyzed for security. It is possible that exiting from mc in some directories with specially constructed names would cause certain commands to be run. Now consider that the middle of very long directories is not shown, and we have a security problem. Also it would be nice if TMPDIR environment variable was supported. If somebody wants the last directory saving to work with csh and tcsh, please fix lib/mc.csh.in from CVS. If nobody replies until next Monday, this functionality will be removed. -- Regards, Pavel Roskin From oskar at osk.mine.nu Thu Oct 31 16:18:27 2002 From: oskar at osk.mine.nu (Oskar Liljeblad) Date: Thu, 31 Oct 2002 17:18:27 +0100 Subject: Need help with csh syntax In-Reply-To: References: Message-ID: <20021031161827.GA745@oskar> On Thursday, October 31, 2002 at 01:11, Pavel Roskin wrote: Hi Pavel! > I have converted mc.sh to the new syntax, when the -P option takes an > argument - the file where the last working directory is placed by mc. An > important fact is that now mc is not guaranteed to create the output file. > It is only created when the panels have been initialized. > > mc.csh needs to be updated as well. I have done simple changes to that > file. However, I'm not experienced in csh enough to do it right. In > particular, the case when the directory contains spaces or some special > symbols (e.g. semicolon) needs to be analyzed for security. Now that the mc.sh file is a script to execute (and not include), is mc.csh really necessary? I mean, all systems that have csh also probably have sh. :) Oskar Liljeblad (oskar at osk.mine.nu) From proski at gnu.org Thu Oct 31 17:03:33 2002 From: proski at gnu.org (Pavel Roskin) Date: Thu, 31 Oct 2002 12:03:33 -0500 (EST) Subject: Need help with csh syntax In-Reply-To: <20021031161827.GA745@oskar> Message-ID: Hello, Oskar! > Now that the mc.sh file is a script to execute (and not include), What? I don't see it on CVS. Can you suggest an implementation? > is mc.csh really necessary? I mean, all systems that have csh also > probably have sh. :) I want csh to change its working directory. At some point it has to run "cd" itself, and I don't want it to be fooled to run anything else. Am I missing something? Actually, I think I have fixed all the problems with the quoting. I still cannot find any csh equivalent for ${TMPDIR-/tmp}. -- Regards, Pavel Roskin From oskar at osk.mine.nu Thu Oct 31 17:11:55 2002 From: oskar at osk.mine.nu (Oskar Liljeblad) Date: Thu, 31 Oct 2002 18:11:55 +0100 Subject: Need help with csh syntax In-Reply-To: References: <20021031161827.GA745@oskar> Message-ID: <20021031171154.GA4301@oskar> On Thursday, October 31, 2002 at 12:33, Pavel Roskin wrote: Hi > > Now that the mc.sh file is a script to execute (and not include), > > What? I don't see it on CVS. Can you suggest an implementation? Oops. My mistake. You said: > > 3. Make mc.sh a wrapper script instead? > > Then you could use alias mc=/usr/lib/mc/bin/mc.sh or whatever. > > I agree. That's what I meant. Perhaps I was unclear. which made me think it would be enough with just that alias... But I forgot the fact that you have to CD.. Oskar (oskar at osk.mine.nu) From samoilov at usb.zp.ua Thu Oct 31 19:10:30 2002 From: samoilov at usb.zp.ua (Andrew V. Samoilov) Date: Thu, 31 Oct 2002 21:10:30 +0200 Subject: mc wish: possibility to configure passive/active ftp mode from UI (fwd) References: <200210311852.g9VIqImF021020@email.zp.ua> Message-ID: <3DC18026.9030408@usb.zp.ua> > ----- Forwarded message from Andrew W. Nosenko ----- > Date: Wed, 16 Oct 2002 12:10:16 +0300 > From: "Andrew W. Nosenko" > Message-ID: <20021016091016.GA28978 at bcs.zp.ua> > To: "Andrew V. Samoilov" > Subject: mc wish: possibility to configure passive/active ftp mode from UI > First part of this patch attached. One more option leaves Special settings and come to UI. Another part should be discussed. I see some ways to implement this feature: 1. Check box in the "FTP: Password required" dialog. Public entries (entries without password) cannot be handled this way. 2. New ftp URL syntax. ChangeLog: * boxes.c (configure_vfs): Ask whether passive connections should be used. -- Regards, Andrew V. Samoilov. -------------- next part -------------- An embedded and charset-unspecified text was scrubbed... Name: boxes.c.patch URL: