2005-05-30; Request for master file as info is invoked.

Dear bug-fixing team,

it happens from time to time, that I want to read AUCTeX's info documentation
as I am working on a TeX document. For that purpose, I do a `C-h i' to get from
the LaTeX buffer to the info buffer. However, XEmacs asks me to set the master
file and suggests, that is located in the info folder of my package
installation (that is: d:\Src\Build\XEmacs\xemacs-packages\info\).

To rule out any misconfiguration, I started up XEmacs with the `--vanilla'
switch and evaluated `(require 'tex-site)' in the scratch buffer. This did not
trigger the odd behavior. After having played with my settings, I found out,
that

  (require 'tex-site)
  (setq-default TeX-master nil)

does. Further testing revealed that the latest beta release of XEmacs
(21.5-b21) does not show that behavior.

I hope, I could provide a concise description of the problem,
Franz.

Emacs  : XEmacs 21.4 (patch 17) "Jumbo Shrimp" [Lucid] (i586-pc-win32) of Fri Apr 22 2005 on PAN
Package: 2005-05-30

current state:
==============
(setq
 AUCTeX-date "2005-05-30"
 window-system 'mswindows

Re: rel-0-9-1; Re: problem with preview-region and TeX-command-region

> > (I don't
> > actually get a dialog box on Windows, just a query in the echo area.)
> 
> You should get one with CVS Emacs in case you use the menu entry for
> "Set Master File" or for previewing a region.

actually, even for the menu invocations on CVS emacs, I don't get a dialog
box, just a query in the echo area. Maybe the emacs build on crasseux.com
was compiled with a preference against this?

> > 1. On Emacs 21.3 (precompiled binary), the dialog doesn't appear at 
all,like
> > you mention. Even when TeX-master is set to nil (query), running
> > TeX-command-region from the menubar doesn't result in a query.
> > (AUCTeX assumes the master file to be `This file'.)
> 
> Even when no _region_.tex file existed before?  Strange.

yeah. I even made sure that (i) both local and global values of
TeX-master were nil, and (ii) the buffer/file in question didn't even
have any \begin{document} or \end{document} that AUCTeX might take a
cue from. When I invoke TeX-command-region from the menu, AUCTeX still
assumes the master file to be that file, and then latex obviously
complains.

What's even more strange, under the same situation as above, when I
use C-c C-r, then AUCTeX asks for a master file, but ignores my answer
and goes ahead with the command. Of course, latex still complains
about a missing \begin{document}, and when I check the _region_.tex
file, there's no mention of the master file I specified, it looks like

Re: 2005-05-30; Request for master file as info is invoked.

* Franz Haeuslschmid (2005-06-01) writes:

> it happens from time to time, that I want to read AUCTeX's info
> documentation as I am working on a TeX document. For that purpose, I
> do a `C-h i' to get from the LaTeX buffer to the info
> buffer. However, XEmacs asks me to set the master file and suggests,
> that is located in the info folder of my package installation (that
> is: d:\Src\Build\XEmacs\xemacs-packages\info\).

I cannot reproduce this with my XEmacs 21.4.17 installation.  Could
you please do a

M-x debug-on-entry RET TeX-master-file-ask RET

before you type `C-h i' and show us the backtrace?  If there is no
backtrace you can try the functions `TeX-master-file' and
`VirTeX-common-initialization' instead of `TeX-master-file-ask'.

--

-- 
Ralf

Re: Re: rel-0-9-1; Re: problem with preview-region and TeX-command-region

* Rajneesh Hegde (2005-06-02) writes:

> actually, even for the menu invocations on CVS emacs, I don't get a dialog
> box, just a query in the echo area. Maybe the emacs build on crasseux.com
> was compiled with a preference against this?

Sorry, but I cannot follow you.  In
<URL:http://article.gmane.org/gmane.emacs.auctex.bugs/137> you wrote
that you got a file dialog.  Now you write that you don't get one.

>> > 1. On Emacs 21.3 (precompiled binary), the dialog doesn't appear at 
> all,like
>> > you mention. Even when TeX-master is set to nil (query), running
>> > TeX-command-region from the menubar doesn't result in a query.
>> > (AUCTeX assumes the master file to be `This file'.)
>> 
>> Even when no _region_.tex file existed before?  Strange.
[...]
> What's even more strange, under the same situation as above, when I
> use C-c C-r, then AUCTeX asks for a master file, but ignores my answer
> and goes ahead with the command. Of course, latex still complains
> about a missing \begin{document}, and when I check the _region_.tex
> file, there's no mention of the master file I specified, it looks like
> AUCTeX disregards my input for the master file.
>
> To reproduce this, use a self-contained latex file (like the example
> you'd given earlier on this thread), create a new latex file that is
> not self-contained, mark aregion in it and then do C-C C-r. When
> AUCTeX queries for a master file, give the other (self-contained) file
> as answer. See if AUCTeX actually uses that answer.

Re: 2005-05-30; Request for master file as info is invoked.

Ralf Angeli <angeli <at> iwi.uni-sb.de> writes:

> * Franz Haeuslschmid (2005-06-01) writes:
>
>> it happens from time to time, that I want to read AUCTeX's info
>> documentation as I am working on a TeX document. For that purpose, I
>> do a `C-h i' to get from the LaTeX buffer to the info
>> buffer. However, XEmacs asks me to set the master file and suggests,
>> that is located in the info folder of my package installation (that
>> is: d:\Src\Build\XEmacs\xemacs-packages\info\).
>
> I cannot reproduce this with my XEmacs 21.4.17 installation.  Could
> you please do a
>
> M-x debug-on-entry RET TeX-master-file-ask RET
>
> before you type `C-h i' and show us the backtrace?

It is attached to this message.

BTW: the output contains control characters.  What is the
preferred way to include such output recordings?

Ralf Angeli <angeli <at> iwi.uni-sb.de> writes:

> * Franz Haeuslschmid (2005-06-01) writes:
>

Re: Re: 2005-05-30; Request for master file as info is invoked.

* Franz Haeuslschmid (2005-06-02) writes:

> Ralf Angeli <angeli <at> iwi.uni-sb.de> writes:
>
>> I cannot reproduce this with my XEmacs 21.4.17 installation.  Could
>> you please do a
>>
>> M-x debug-on-entry RET TeX-master-file-ask RET
>>
>> before you type `C-h i' and show us the backtrace?
>
> It is attached to this message.

Okay, thanks.  Could you please also show us the output of
C-h v find-file-hooks RET
after doing `C-h i' (that means with the info buffer being active) as
well as in a completely different buffer, say the *scratch* buffer.

In addition it would be interesting to see the output of
C-h v auto-mode-alist RET

> BTW: the output contains control characters.  What is the
> preferred way to include such output recordings?

That depends.  If you know this stuff is not important you can remove
those characters (Gnus can do this automatically).  But the safe
option is to send it as an attachment as you did.

--

-- 
Ralf

Re: Re: 2005-05-30; Request for master file as info is invoked.

Franz Haeuslschmid <fhaeuslschmid <at> t-online.de> writes:

> Ralf Angeli <angeli <at> iwi.uni-sb.de> writes:
>>
>> M-x debug-on-entry RET TeX-master-file-ask RET
>>
>> before you type `C-h i' and show us the backtrace?
>
> It is attached to this message.
>
> BTW: the output contains control characters.  What is the
> preferred way to include such output recordings?

If they are of relevance, an attachment like you did should be ok.
However, I don't think anybody ever actually disassembled the byte
codes from a backtrace, so in this case it would have been slightly
more convenient to just paste the buffer in the text message, and when
Emacs asks what it should do about control characters in the message,
replace them (r) with question marks.

But the attachment is, though slightly less convenient, the safe way
if you don't know whether every byte might be important.

--

-- 
David Kastrup, Kriemhildstr. 15, 44793 Bochum

Re: 2005-05-30; Request for master file as info is invoked.

Ralf Angeli <angeli <at> iwi.uni-sb.de> writes:

> * Franz Haeuslschmid (2005-06-02) writes:
>
>> Ralf Angeli <angeli <at> iwi.uni-sb.de> writes:
>>
>>> I cannot reproduce this with my XEmacs 21.4.17 installation.  Could
>>> you please do a
>>>
>>> M-x debug-on-entry RET TeX-master-file-ask RET
>>>
>>> before you type `C-h i' and show us the backtrace?
>>
>> It is attached to this message.
>
> Okay, thanks.  Could you please also show us the output of
> C-h v find-file-hooks RET
> after doing `C-h i' (that means with the info buffer being active) as
> well as in a completely different buffer, say the *scratch* buffer.

With active info buffer (documentation is stripped away):

`find-file-hooks' is a variable declared in Lisp.
  -- loaded from "d:\Src\xemacs-21.4\lisp\files.elc"

Value: (#<compiled-function nil "...(21)" [TeX-master file-exists-p buffer-file-name shared
TeX-master-file nil t TeX-update-style] 4> font-lock-set-defaults)

Then switching to the *scratch* buffer

Re: Re: 2005-05-30; Request for master file as info is invoked.

* Franz Haeuslschmid (2005-06-02) writes:

> Ralf Angeli <angeli <at> iwi.uni-sb.de> writes:
>
>> Okay, thanks.  Could you please also show us the output of
>> C-h v find-file-hooks RET
>> after doing `C-h i' (that means with the info buffer being active) as
>> well as in a completely different buffer, say the *scratch* buffer.
>
> With active info buffer (documentation is stripped away):
[...]
> Value: (#<compiled-function nil "...(21)" [TeX-master file-exists-p buffer-file-name shared
TeX-master-file nil t TeX-update-style] 4> font-lock-set-defaults)
[...]
> Then switching to the *scratch* buffer
[...]
> Value: (#<compiled-function nil "...(21)" [TeX-master file-exists-p buffer-file-name shared
TeX-master-file nil t TeX-update-style] 4> font-lock-set-defaults)

This is a bug in XEmacs.  It fails to make the hook buffer-local.  I
sent a bug report to xemacs-beta and added a preliminary patch to
AUCTeX which helps XEmacs a bit accomplishing its job.  It would be
nice if you could tell us if this fixes your problem.

(Interestingly after debugging this, the error showed up here as
well.  Don't ask me why I could not observe it before.)

--

-- 
Ralf

Re: 2005-05-30; Request for master file as info is invoked.

Ralf Angeli <angeli <at> iwi.uni-sb.de> writes:

> * Franz Haeuslschmid (2005-06-02) writes:
>
>> Ralf Angeli <angeli <at> iwi.uni-sb.de> writes:
>>
>>> Okay, thanks.  Could you please also show us the output of
>>> C-h v find-file-hooks RET
>>> after doing `C-h i' (that means with the info buffer being active) as
>>> well as in a completely different buffer, say the *scratch* buffer.
>>
>> With active info buffer (documentation is stripped away):
> [...]
>> Value: (#<compiled-function nil "...(21)" [TeX-master file-exists-p buffer-file-name shared
TeX-master-file nil t TeX-update-style] 4> font-lock-set-defaults)
> [...]
>> Then switching to the *scratch* buffer
> [...]
>> Value: (#<compiled-function nil "...(21)" [TeX-master file-exists-p buffer-file-name shared
TeX-master-file nil t TeX-update-style] 4> font-lock-set-defaults)
>
> This is a bug in XEmacs.  It fails to make the hook buffer-local.  I
> sent a bug report to xemacs-beta and added a preliminary patch to
> AUCTeX which helps XEmacs a bit accomplishing its job.  It would be
> nice if you could tell us if this fixes your problem.

Yes, it does indeed fix the problem.  Thank you very much.

> (Interestingly after debugging this, the error showed up here as
> well.  Don't ask me why I could not observe it before.)