More problems on Cygwin with the subshell - was: Re: [PATCH suggestion] exceptions.cc, interrupt_setup ()

[Pavel Tsekov]

PT> I suggest the following very simple patch. Since I may have not
PT> understand all the specifics of the signal handling mechanism I offer
PT> it for discussion. Just for the record - this patch solves that
PT> outstanding problem with MC.

This patch has been checked in: http://www.cygwin.com/ml/cygwin-patches/2002-q3/msg00294.html

For those who are testing the Cygwin version of MC, I suggest to get the latest
snapshot dll or build from source (http://www.cygwin.com/snapshots/).

Thanks to Pavel Rosking for his help! :)

Now the bad news:

  There still seems to be a lockup on Cygwin when using the subshell.
  It appears not so often but it's there :(

  Right now I'm debugging this - it very much looks like what Pavel
  Roskin suggested in his post: http://mail.gnome.org/archives/mc-devel/2002-August/msg00040.html
  It seems like the subshell is not getting the SIGCONT and
  additionally MC is not getting the SIGCHLD. But I won't speculate more,
  before I have solid  evidence. What I know for sure is that at the
  time of the lockup the subshell hangs in the Cygwin handler for the
  SIGSTOP signal i.e. it is suspended, and MC hangs in the select()
  called by feed_subshell () which is not interrupted (by a SIGCHLD).

  Expect more on that issue ... soon :)

Thanks! :)

Pavel Tsekov