Re: Emacs-diffs Digest, Vol 74, Issue 135

Re: Emacs-diffs Digest, Vol 74, Issue 135

[Eli Zaretskii]

> Date: Thu, 29 Jan 2009 09:35:18 -0500
> From: emacs-diffs-request@gnu.org
> 
> 	* blockinput.h: Add declaration for pending_signals.

I don't see this change installed.  Did you forget or is the log entry
wrong?

As things are now, the only header that declares pending_signals is
lisp.h, and it does that only when SYNC_INPUT is defined.  However,
pending_signals is used even when SYNC_INPUT is not defined, so I
wonder whether you forgot to commit blockinput.h or all uses of
pending_signals should be moved inside #ifdef SYNC_INPUT.

This currently breaks the MSDOS build.

Re: Emacs-diffs Digest, Vol 74, Issue 135

[Eli Zaretskii]

> Date: Thu, 29 Jan 2009 21:20:36 +0200
> From: Eli Zaretskii <eliz@gnu.org>
> Cc: emacs-devel@gnu.org

Yidong, would you please respond to the below?  I want to fix the DOS
build before the pretest starts, but it's currently broken because
blockinput.h (or any other header) doesn't declare pending_signals,
except conditioned by SYNC_INPUT.

Please help me fix this.

> > Date: Thu, 29 Jan 2009 09:35:18 -0500
> > From: emacs-diffs-request@gnu.org
> > 
> > 	* blockinput.h: Add declaration for pending_signals.
> 
> I don't see this change installed.  Did you forget or is the log entry
> wrong?
> 
> As things are now, the only header that declares pending_signals is
> lisp.h, and it does that only when SYNC_INPUT is defined.  However,
> pending_signals is used even when SYNC_INPUT is not defined, so I
> wonder whether you forgot to commit blockinput.h or all uses of
> pending_signals should be moved inside #ifdef SYNC_INPUT.
> 
> This currently breaks the MSDOS build.
> 
> 
> 

Re: Emacs-diffs Digest, Vol 74, Issue 135

[Chong Yidong]

Eli Zaretskii <eliz@gnu.org> writes:

> Yidong, would you please respond to the below?  I want to fix the DOS
> build before the pretest starts, but it's currently broken because
> blockinput.h (or any other header) doesn't declare pending_signals,
> except conditioned by SYNC_INPUT.
>
> Please help me fix this.

Juanma has a patch putting pending_signals in #ifdef SYNC_INPUT, and
I've just asked him to commit it.

Re: Emacs-diffs Digest, Vol 74, Issue 135

[Juanma Barranquero]

On Fri, Jan 30, 2009 at 13:38, Chong Yidong <cyd@stupidchicken.com> wrote:

> Juanma has a patch putting pending_signals in #ifdef SYNC_INPUT, and
> I've just asked him to commit it.

That said, I think that's ugly as hell (I've committed it so Emacs
compiles again).

pending_signals could be defined in lisp.h outside the "#ifdef
SYNC_INPUT", which would save a lot of #ifdefs, with the only side
effect that the variable would be assigned in several situations, but
not used (on non-SYNC_INPUT builds). That's not pretty either, but I
think that's what's done with interrupt_input_pending.

Or we could use a macro to set pending_signals, which would do nothing
in the relevant case.

    Juanma

Re: Emacs-diffs Digest, Vol 74, Issue 135

[Eli Zaretskii]

> Date: Fri, 30 Jan 2009 14:14:26 +0100
> From: Juanma Barranquero <lekktu@gmail.com>
> Cc: Eli Zaretskii <eliz@gnu.org>, emacs-devel@gnu.org
> 
> On Fri, Jan 30, 2009 at 13:38, Chong Yidong <cyd@stupidchicken.com> wrote:
> 
> > Juanma has a patch putting pending_signals in #ifdef SYNC_INPUT, and
> > I've just asked him to commit it.
> 
> That said, I think that's ugly as hell (I've committed it so Emacs
> compiles again).

Thanks.

> pending_signals could be defined in lisp.h outside the "#ifdef
> SYNC_INPUT", which would save a lot of #ifdefs

Some of the #ifdef's are in w32*.c files.  Why did you need that?
Isn't it known whether the Windows port uses or does not use
SYNC_INPUT?

Re: Emacs-diffs Digest, Vol 74, Issue 135

[Juanma Barranquero]

On Fri, Jan 30, 2009 at 16:53, Eli Zaretskii <eliz@gnu.org> wrote:

> Some of the #ifdef's are in w32*.c files.  Why did you need that?
> Isn't it known whether the Windows port uses or does not use
> SYNC_INPUT?

It does not use it.

But, what would you do? Just remove the #ifdef'd code?

    Juanma

Re: Emacs-diffs Digest, Vol 74, Issue 135

[Eli Zaretskii]

> Date: Fri, 30 Jan 2009 17:59:08 +0100
> From: Juanma Barranquero <lekktu@gmail.com>
> Cc: cyd@stupidchicken.com, emacs-devel@gnu.org
> 
> On Fri, Jan 30, 2009 at 16:53, Eli Zaretskii <eliz@gnu.org> wrote:
> 
> > Some of the #ifdef's are in w32*.c files.  Why did you need that?
> > Isn't it known whether the Windows port uses or does not use
> > SYNC_INPUT?
> 
> It does not use it.
> 
> But, what would you do? Just remove the #ifdef'd code?

Yes.  Is there a reason not to?

Re: Emacs-diffs Digest, Vol 74, Issue 135

[Juanma Barranquero]

On Fri, Jan 30, 2009 at 18:41, Eli Zaretskii <eliz@gnu.org> wrote:

> Yes.  Is there a reason not to?

I assumed that Chong had added that code for a reason ;-)

OK, I'll remove these bits. I suppose if the Windows port were to be
changed to use SYNC_INPUT we'd notice.

    Juanma

Re: Emacs-diffs Digest, Vol 74, Issue 135

[Stefan Monnier]

> That's not pretty either, but I think that's what's done with
> interrupt_input_pending.

Actually, interrupt_input_pending (just like pending_atimers) is used
even when SYNC_INPUT is not used: it's used to keep track of pending
signals that happened during BLOCKINPUT.


        Stefan

Re: Emacs-diffs Digest, Vol 74, Issue 135

[Eli Zaretskii]

> From: Chong Yidong <cyd@stupidchicken.com>
> Date: Fri, 30 Jan 2009 07:38:59 -0500
> Cc: emacs-devel@gnu.org
> 
> Eli Zaretskii <eliz@gnu.org> writes:
> 
> > Yidong, would you please respond to the below?  I want to fix the DOS
> > build before the pretest starts, but it's currently broken because
> > blockinput.h (or any other header) doesn't declare pending_signals,
> > except conditioned by SYNC_INPUT.
> >
> > Please help me fix this.
> 
> Juanma has a patch putting pending_signals in #ifdef SYNC_INPUT, and
> I've just asked him to commit it.

Thanks.