emacs 25 broken?

emacs 25 broken?

[Stephen Leake]

Whenever I try to use emacs 25, something doesn't work right, and when
trying to debug it I always run into bug #19611; "e (current-buffer)" in
edebug returns the wrong buffer.

Does anyone else see this problem? It's blocking all my work on master.

-- 
-- Stephe

Re: emacs 25 broken?

[Eli Zaretskii]

> From: Stephen Leake <stephen_leake@stephe-leake.org>
> Date: Tue, 24 Feb 2015 11:46:32 -0600
> 
> Whenever I try to use emacs 25, something doesn't work right, and when
> trying to debug it I always run into bug #19611; "e (current-buffer)" in
> edebug returns the wrong buffer.
> 
> Does anyone else see this problem? It's blocking all my work on master.

Why do you need "e (current-buffer)" in Edebug so badly that it blocks
your work?

Re: emacs 25 broken?

[Stephen Leake]

Eli Zaretskii <eliz@gnu.org> writes:

>> From: Stephen Leake <stephen_leake@stephe-leake.org>
>> Date: Tue, 24 Feb 2015 11:46:32 -0600
>> 
>> Whenever I try to use emacs 25, something doesn't work right, and when
>> trying to debug it I always run into bug #19611; "e (current-buffer)" in
>> edebug returns the wrong buffer.
>> 
>> Does anyone else see this problem? It's blocking all my work on master.
>
> Why do you need "e (current-buffer)" in Edebug so badly that it blocks
> your work?

Seriously? If I can't trust the debugger, how am I supposed to make
progress?

The real bug (in ada-mode) I'm trying to fix is in code that switches
buffers. (That bug does not exist when the same code is run in emacs
24). Any expression I evaluate that relies on the current buffer is
wrong in the debugger, so it's impossible to figure out what's going on.

I did not record in the bug that `with-current-buffer' is not broken;
executing M-x sal-debug gives the correct results.

Any suggestions for further investigation?

-- 
-- Stephe

Re: emacs 25 broken?

[Stephen Leake]

Stephen Leake <stephen_leake@stephe-leake.org> writes:

> Eli Zaretskii <eliz@gnu.org> writes:
>
>>> From: Stephen Leake <stephen_leake@stephe-leake.org>
>>> Date: Tue, 24 Feb 2015 11:46:32 -0600
>>> 
>>> Whenever I try to use emacs 25, something doesn't work right, and when
>>> trying to debug it I always run into bug #19611; "e (current-buffer)" in
>>> edebug returns the wrong buffer.
>>> 
>>> Does anyone else see this problem? It's blocking all my work on master.
>>
>> Why do you need "e (current-buffer)" in Edebug so badly that it blocks
>> your work?
>
> Seriously? If I can't trust the debugger, how am I supposed to make
> progress?
>
> The real bug (in ada-mode) I'm trying to fix is in code that switches
> buffers. (That bug does not exist when the same code is run in emacs
> 24). Any expression I evaluate that relies on the current buffer is
> wrong in the debugger, so it's impossible to figure out what's going on.
>
> I did not record in the bug that `with-current-buffer' is not broken;
> executing M-x sal-debug gives the correct results.
>
> Any suggestions for further investigation?

I just noticed on the bug report that Andreas Politz isolated the commit
that broke it: commit 6194477a622ce9293162d16a7ba98c9cfc18d124 by Stefan
(I'm guessing Andreas ran git-bisect; I should have tried that).

That's a fairly small change to src/eval.c; reverting that commit does
fix bug #19611, and lets me make progress on my ada-mode bug.

I don't understand the change in eval.c, so I don't know how to fix it.

-- 
-- Stephe

Re: emacs 25 broken?

[Artur Malabarba]

You mentioned using edebug. Having you tried exploring the bug with
`(debug)' or `debug-on-entry' instead?

2015-02-24 21:15 GMT+00:00 Stephen Leake <stephen_leake@stephe-leake.org>:
> Eli Zaretskii <eliz@gnu.org> writes:
>
>>> From: Stephen Leake <stephen_leake@stephe-leake.org>
>>> Date: Tue, 24 Feb 2015 11:46:32 -0600
>>>
>>> Whenever I try to use emacs 25, something doesn't work right, and when
>>> trying to debug it I always run into bug #19611; "e (current-buffer)" in
>>> edebug returns the wrong buffer.
>>>
>>> Does anyone else see this problem? It's blocking all my work on master.
>>
>> Why do you need "e (current-buffer)" in Edebug so badly that it blocks
>> your work?
>
> Seriously? If I can't trust the debugger, how am I supposed to make
> progress?
>
> The real bug (in ada-mode) I'm trying to fix is in code that switches
> buffers. (That bug does not exist when the same code is run in emacs
> 24). Any expression I evaluate that relies on the current buffer is
> wrong in the debugger, so it's impossible to figure out what's going on.
>
> I did not record in the bug that `with-current-buffer' is not broken;
> executing M-x sal-debug gives the correct results.
>
> Any suggestions for further investigation?
>
> --
> -- Stephe
>

Re: emacs 25 broken?

[Stefan Monnier]

> I don't understand the change in eval.c, so I don't know how to fix it.

IIUC the problem is not in eval.c but in edebug.el, but I haven't found
the time to dig into it.

What the eval.c change does is that backtrace-eval now doesn't only
restore the variable bindings to what they were in the relevant stack
frame, but also restores the current-buffer (and point).

It looks like Edebug evaluates its code in a frame where current-buffer
is wrong or something like that, or tries to force the use of the right
buffer "by hand" and this ends up backfiring with the new system which
tries to do it "automatically".


        Stefan

Re: emacs 25 broken?

[Stefan Monnier]

> I don't understand the change in eval.c, so I don't know how to fix it.

Of course, maybe this change is indeed broken.


        Stefan

Re: emacs 25 broken?

[Eli Zaretskii]

> From: Stephen Leake <stephen_leake@stephe-leake.org>
> Date: Tue, 24 Feb 2015 15:15:00 -0600
> 
> Eli Zaretskii <eliz@gnu.org> writes:
> 
> >> From: Stephen Leake <stephen_leake@stephe-leake.org>
> >> Date: Tue, 24 Feb 2015 11:46:32 -0600
> >> 
> >> Whenever I try to use emacs 25, something doesn't work right, and when
> >> trying to debug it I always run into bug #19611; "e (current-buffer)" in
> >> edebug returns the wrong buffer.
> >> 
> >> Does anyone else see this problem? It's blocking all my work on master.
> >
> > Why do you need "e (current-buffer)" in Edebug so badly that it blocks
> > your work?
> 
> Seriously? If I can't trust the debugger, how am I supposed to make
> progress?

There's a large gap between "e (current-buffer)" shows something other
than what you expect and "I cannot trust the debugger".  There's
nothing wrong with Edebug in general, so I don't see how that
particular issue can block any work.  FWIW, I use Edebug every day
(and yes, I do see the same issue with current-buffer).

> The real bug (in ada-mode) I'm trying to fix is in code that switches
> buffers. (That bug does not exist when the same code is run in emacs
> 24). Any expression I evaluate that relies on the current buffer is
> wrong in the debugger, so it's impossible to figure out what's going on.

You can always work around this.  E.g., add a variable that gets
assigned the buffer and look at it.  Or even add a 'message' call that
prints it.

> I did not record in the bug that `with-current-buffer' is not broken;
> executing M-x sal-debug gives the correct results.

There you go.

Re: emacs 25 broken?

[Richard Stallman]

[[[ To any NSA and FBI agents reading my email: please consider    ]]]
[[[ whether defending the US Constitution against all enemies,     ]]]
[[[ foreign or domestic, requires you to follow Snowden's example. ]]]

I often do e (current-buffer) in the Emacs debugger because I am debugging
Rmail, and the current buffer is often not what's displayed in the window.
I need to make sure of which buffer is really being operated on.
A wrong value from e (current-buffer) would cause terrible confusion.

I use the ordinary debugger, not edebug mode, but the issue is the same
for both.

So I think this bug is important, and the right thing to do is fix it.

-- 
Dr Richard Stallman
President, Free Software Foundation
51 Franklin St
Boston MA 02110
USA
www.fsf.org  www.gnu.org
Skype: No way! See stallman.org/skype.html.