23.0.60; OSX: isatty

23.0.60; OSX: isatty

[Markus Triska]

This narrows down the report regarding accept-process-output a bit: I've
built Emacs CVS against X11 only (no Carbon) on OSX 10.4.10. The
following C program provides a simple test case:

   #include <stdio.h>
   #include <unistd.h>
   int main() { printf("isatty(stdin): %d\n", isatty(fileno(stdin)));
                fflush(stdout); return 0; }

When I run this program from within Emacs with M-x comint-run, I get
different outputs in the buffer, depending on the Emacs version:

   *) Before multi-tty merge:

      isatty(stdin): 1

   *) After multi-tty merge (current CVS):

      isatty(stdin): 0

For this reason, some programs stop showing prompts. M-x shell, for
example, does not show any prompt, but the (bash) shell is there, and I
can enter and execute commands in the buffer.


In GNU Emacs 23.0.60.1 (i386-apple-darwin8.10.1, X toolkit)
 of 2007-11-12 on mt-computer.local
Windowing system distributor `The X.Org Foundation', version 11.0.60802000
configured using `configure  '--without-carbon' '--with-x' '--without-gif' '--without-tiff' '--enable-font-backend' 'LIBS=-lresolv''

Re: 23.0.60; OSX: isatty

[Stefan Monnier]

> This narrows down the report regarding accept-process-output a bit: I've
> built Emacs CVS against X11 only (no Carbon) on OSX 10.4.10. The
> following C program provides a simple test case:

>    #include <stdio.h>
>    #include <unistd.h>
>    int main() { printf("isatty(stdin): %d\n", isatty(fileno(stdin)));
>                 fflush(stdout); return 0; }

> When I run this program from within Emacs with M-x comint-run, I get
> different outputs in the buffer, depending on the Emacs version:

>    *) Before multi-tty merge:

>       isatty(stdin): 1

>    *) After multi-tty merge (current CVS):

>       isatty(stdin): 0

> For this reason, some programs stop showing prompts. M-x shell, for
> example, does not show any prompt, but the (bash) shell is there, and I
> can enter and execute commands in the buffer.

What's the value of `process-connection-type'?


        Stefan

Re: 23.0.60; OSX: isatty

[Markus Triska]

Stefan Monnier <monnier@iro.umontreal.ca> writes:

> What's the value of `process-connection-type'?

It is nil by default (emacs -q). Manually setting it to t seems to
make everything work. Thank you very much!

Re: 23.0.60; OSX: isatty

[Stefan Monnier]

>> What's the value of `process-connection-type'?

> It is nil by default (emacs -q).

Please try it with "emacs -Q".  If it's still nil, then try and figure
where this is coming from.


        Stefan

Re: 23.0.60; OSX: isatty

[Markus Triska]

Stefan Monnier <monnier@iro.umontreal.ca> writes:

> Please try it with "emacs -Q".  If it's still nil, then try and figure
> where this is coming from.

It also happens with -Q. It's from init_process (process.c, l. 7152ff):

#if defined (DARWIN) || defined (MAC_OSX)
  /* PTYs are broken on Darwin < 6, but are sometimes useful for interactive
     processes.  As such, we only change the default value.  */
 if (initialized)
  {
    char *release = get_operating_system_release();
    if (!release || !release[0] || (release[0] < MIN_PTY_KERNEL_VERSION
				    && release[1] == '.')) {
      Vprocess_connection_type = Qnil;
    }
  }
#endif

This is executed before init_editfns has had a chance to initialize
Voperating_system_release (MAC_OSX is defined on my platform).

Re: 23.0.60; OSX: isatty

[Markus Triska]

Stefan Monnier <monnier@iro.umontreal.ca> writes:

> figure where this is coming from.

For completeness, this is the relevant commit:

commit 6c194ef1b42ba72670dbfaa82b7eacffbf22e2d4
Author: Karoly Lorentey <lorentey@elte.hu>
Date:   Mon Mar 29 12:05:03 2004 +0000

    Fix initialization order at bootstrap (Dan Nicolaescu).
    
    src/emacs.c (main): Call init_process before init_display to prevent
    losing the keyboard on stdin (reported by Dan Nicolaescu).
    git-archimport-id: lorentey@elte.hu--2004/emacs--multi-tty--0--patch-128

diff --git a/src/emacs.c b/src/emacs.c
index 43d2f85..1c7d595 100644
--- a/src/emacs.c
+++ b/src/emacs.c
@@ -1578,6 +1578,7 @@ main (argc, argv
 #endif  /* end #ifdef HAVE_NTGUI */
     }
 
+  init_process (); /* init_display uses add_keyboard_wait_descriptor. */
 #ifndef MAC_OS8
   /* Called before init_window_once for Mac OS Classic.  */
   init_keyboard ();	/* This too must precede init_sys_modes.  */
@@ -1604,7 +1605,6 @@ main (argc, argv
 #ifdef VMS
   init_vmsfns ();
 #endif /* VMS */
-  init_process ();
 #ifdef HAVE_SOUND
   init_sound ();
 #endif

[patch] Re: 23.0.60; OSX: isatty

[Markus Triska]

Stefan Monnier <monnier@iro.umontreal.ca> writes:

> where this is coming from.

The following patch closes the issue.


2007-11-19  Markus Triska  <markus.triska@gmx.at>

	* emacs.c (main): call init_editfns before init_process, since
	init_process sets Vprocess_connection_type depending on OS release


diff --git a/src/emacs.c b/src/emacs.c
index 8a51b8c..af6a7b7 100644
--- a/src/emacs.c
+++ b/src/emacs.c
@@ -1666,6 +1666,7 @@ main (argc, argv
 #endif  /* HAVE_NTGUI */
     }
 
+  init_editfns (); /* init_process uses Voperating_system_release. */
   init_process (); /* init_display uses add_keyboard_wait_descriptor. */
 #ifndef MAC_OS8
   /* Called before init_window_once for Mac OS Classic.  */
@@ -1688,7 +1689,6 @@ main (argc, argv
   init_image ();
 #endif /* HAVE_WINDOW_SYSTEM */
   init_macros ();
-  init_editfns ();
   init_floatfns ();
 #ifdef VMS
   init_vmsfns ();

Re: [patch] Re: 23.0.60; OSX: isatty

[Stefan Monnier]

> The following patch closes the issue.


> 2007-11-19  Markus Triska  <markus.triska@gmx.at>

> 	* emacs.c (main): call init_editfns before init_process, since
> 	init_process sets Vprocess_connection_type depending on OS release

Installed.  Thanks for your help.  Sorry for the delay,


        Stefan