commends in Python mode

commends in Python mode

[Paul Pogonyshev]

Try inputting the following in a Python mode buffer:

test = (1,
        # Apostrophe: '
        2)

You can see that the apostrophe breaks sexp scanning (the closing
paren is not paired with the opening) and probably more.  This
doesn't happen in e.g. C-mode, so it is a problem of Python mode.

Paul

Re: commends in Python mode

[Steven Huwig]

On Dec 25, 2006, at 5:30 PM, Paul Pogonyshev wrote:

> Try inputting the following in a Python mode buffer:
>
> test = (1,
>         # Apostrophe: '
>         2)
>
> You can see that the apostrophe breaks sexp scanning (the closing
> paren is not paired with the opening) and probably more.  This
> doesn't happen in e.g. C-mode, so it is a problem of Python mode.
>
> Paul

This example works for me in a recent build from CVS. Are you
using python.el from Emacs 22, or the old python-mode.el?


-- Steve

Re: comments in Python mode

[Paul Pogonyshev]

Steven Huwig wrote:
> 
> On Dec 25, 2006, at 5:30 PM, Paul Pogonyshev wrote:
> 
> > Try inputting the following in a Python mode buffer:
> >
> > test = (1,
> >         # Apostrophe: '
> >         2)
> >
> > You can see that the apostrophe breaks sexp scanning (the closing
> > paren is not paired with the opening) and probably more.  This
> > doesn't happen in e.g. C-mode, so it is a problem of Python mode.
> >
> > Paul
> 
> This example works for me in a recent build from CVS. Are you
> using python.el from Emacs 22, or the old python-mode.el?

Weird, because I just tested and sexp scanning is still broken.  I
updated from CVS, opened `python.el' and invoked `eval-buffer'.  No
change.

Paul

Re: comments in Python mode

[Steven Huwig]

On Dec 27, 2006, at 12:07 PM, Paul Pogonyshev wrote:

> Steven Huwig wrote:
>>
>> On Dec 25, 2006, at 5:30 PM, Paul Pogonyshev wrote:
>>
>>> Try inputting the following in a Python mode buffer:
>>>
>>> test = (1,
>>>         # Apostrophe: '
>>>         2)
>>>
>>> You can see that the apostrophe breaks sexp scanning (the closing
>>> paren is not paired with the opening) and probably more.  This
>>> doesn't happen in e.g. C-mode, so it is a problem of Python mode.
>>>
>>> Paul
>>
>> This example works for me in a recent build from CVS. Are you
>> using python.el from Emacs 22, or the old python-mode.el?
>
> Weird, because I just tested and sexp scanning is still broken.  I
> updated from CVS, opened `python.el' and invoked `eval-buffer'.  No
> change.


Maybe we are looking at different behavior.  I checked that
show-paren-mode and forward- and backward-sexp worked.  Is that what
you are doing?

I will update my Emacs to the current CVS version; my working build is
a few weeks old.  Maybe it is a regression...

-- Steve

Re: comments in Python mode

[Chris Moore]

Steven Huwig <steven.huwig@gmail.com> writes:

> Maybe we are looking at different behavior.  I checked that
> show-paren-mode and forward- and backward-sexp worked.  Is that what
> you are doing?

It's what I'm doing, and it fails for me, using python.el revision
1.52, which hasn't been changed for a month or so.

I've tried lots of older versions of python.el, but they all (1.1,
1.10, 1.20, 1.30, 1.40, 1.49 - 1.52) exhibit the same problem.  I
guess the problem is in some other source file.

Here are the backtraces I get:

Debugger entered--Lisp error: (scan-error "Unbalanced parentheses" 8 47)
  scan-sexps(8 1)
  forward-sexp(1)
  call-interactively(forward-sexp)

Debugger entered--Lisp error: (scan-error "Unbalanced parentheses" 45 1)
  scan-sexps(46 -1)
  forward-sexp(-1)
  backward-sexp(1)
  call-interactively(backward-sexp)

Chris.

Re: comments in Python mode

[Steven Huwig]

On Jan 3, 2007, at 7:08 PM, Chris Moore wrote:

> Steven Huwig <steven.huwig@gmail.com> writes:
>
>> Maybe we are looking at different behavior.  I checked that
>> show-paren-mode and forward- and backward-sexp worked.  Is that what
>> you are doing?
>
> It's what I'm doing, and it fails for me, using python.el revision
> 1.52, which hasn't been changed for a month or so.
>
> I've tried lots of older versions of python.el, but they all (1.1,
> 1.10, 1.20, 1.30, 1.40, 1.49 - 1.52) exhibit the same problem.  I
> guess the problem is in some other source file.

Well, I rebuilt my emacs from CVS on Jan 1, and it still works fine
for me.  This is quite strange.

Is there anyone else *besides* me that can input

test = (1,
         # Apostrophe: '
         2)

into a Python buffer and have the sexp-based commands work correctly?

-- Steve

Re: comments in Python mode

[Paul Pogonyshev]

Steven Huwig wrote:
> Well, I rebuilt my emacs from CVS on Jan 1, and it still works fine
> for me.  This is quite strange.
> 
> Is there anyone else *besides* me that can input
> 
> test = (1,
>          # Apostrophe: '
>          2)
> 
> into a Python buffer and have the sexp-based commands work correctly?

Maybe you have customized something.  Try running `emacs -Q' (gives
the same incorrect behavior here with sources of today CVS.)

Paul

Re: comments in Python mode

[Steven Huwig]

On Jan 4, 2007, at 7:11 PM, Paul Pogonyshev wrote:

> Steven Huwig wrote:
>> Well, I rebuilt my emacs from CVS on Jan 1, and it still works fine
>> for me.  This is quite strange.
>>
>> Is there anyone else *besides* me that can input
>>
>> test = (1,
>>          # Apostrophe: '
>>          2)
>>
>> into a Python buffer and have the sexp-based commands work correctly?
>
> Maybe you have customized something.  Try running `emacs -Q' (gives
> the same incorrect behavior here with sources of today CVS.)
>
> Paul

Of course that was the problem.  Sorry for the confusion.  I load CEDET
and it probably handles the case better.

Too much holiday food befuddled my brain, maybe. :-)

-- Steve

Re: commends in Python mode

[Stefan Monnier]

> Try inputting the following in a Python mode buffer:
> test = (1,
>         # Apostrophe: '
>         2)

> You can see that the apostrophe breaks sexp scanning (the closing
> paren is not paired with the opening) and probably more.  This
> doesn't happen in e.g. C-mode, so it is a problem of Python mode.

I believe you're just suffering the consequences of having
parse-sexp-ignore-comments set to nil.  IMO it should always be set to t,
but for some reason it defaults to nil.


        Stefan

Re: commends in Python mode

[Chris Moore]

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

> I believe you're just suffering the consequences of having
> parse-sexp-ignore-comments set to nil.  IMO it should always be set to t,
> but for some reason it defaults to nil.

Setting parse-sexp-ignore-comments to t fixes the bug for me.

Is there a good reason why it defaults to nil, or is this another of
the things we're not allowed to talk about?

Re: commends in Python mode

[Paul Pogonyshev]

Chris Moore wrote:
> Stefan Monnier <monnier@iro.umontreal.ca> writes:
> 
> > I believe you're just suffering the consequences of having
> > parse-sexp-ignore-comments set to nil.  IMO it should always be set to t,
> > but for some reason it defaults to nil.
> 
> Setting parse-sexp-ignore-comments to t fixes the bug for me.
> 
> Is there a good reason why it defaults to nil, or is this another of
> the things we're not allowed to talk about?

Glad the source of the problem is clear now.  It makes to set default
to t to me too, but if it is not done, let's make Python mode at least
set it in its buffers.

Paul

Re: commends in Python mode

[Richard Stallman]

    Glad the source of the problem is clear now.  It makes to set default
    to t to me too, but if it is not done, let's make Python mode at least
    set it in its buffers.

If people are sure this gives good results in Python mode, I am not
against it.  To change the default is too big for right now, but a
change limited to Python mode is small enough to be ok if Python
users find it good.

Re: commends in Python mode

[Richard Stallman]

    I believe you're just suffering the consequences of having
    parse-sexp-ignore-comments set to nil.  IMO it should always be set to t,
    but for some reason it defaults to nil.

After the release, we could try changing the default to t.