End of file during parsing, single newsgroup

End of file during parsing, single newsgroup

[Stephen Harker]

I am using GNU Emacs 22.2.1 (powerpc-redhat-linux-gnu, GTK+ Version
2.10.4) with gnus v5.11.  Starting about two days ago I have suddenly
started having a problem with one newsgroup I am subscribed before.
The relevant messages are:

Retrieving newsgroup: comp.text.tex...
Fetching headers for comp.text.tex...done
Scoring...done
Generating summary...done
Auto-saving...
Wrapped [5 times]
No more unread articles
No more unread articles (selecting rec.arts.sf.written)...
Retrieving newsgroup: rec.arts.sf.written...
Fetching headers for rec.arts.sf.written...
byte-code: End of file during parsing

I have no problem with any other newsgroup.  I have not changed my
.emacs in months.  As can be seen the previous newsgroup reads without
problems.  I tried removing my ADAPT and SCORE files as these might
have been a source of problems, but it makes no difference.  I have
marked the whole group read, but the problem persists with new posts.
Nothing appears to have changed in /usr/share/emacs/.

I would appreciate any suggestions.

-- 
Stephen Harker                           s.harker@adfa.edu.au
PEMS                http://sjharker.customer.netspace.net.au/
UNSW@ADFA

Re: End of file during parsing, single newsgroup

[Tassilo Horn]

Stephen Harker <sjharker@netspace.net.au> writes:

Hi Stephen,

> Retrieving newsgroup: rec.arts.sf.written...
> Fetching headers for rec.arts.sf.written...
> byte-code: End of file during parsing

Before trying to enter that group, do M-x toggle-debug-on-error RET and
post the backtrace here.

Bye,
Tassilo

Re: End of file during parsing, single newsgroup

[Stephen Harker]

Tassilo Horn <tsdh@gnu.org> writes:

> Stephen Harker <sjharker@netspace.net.au> writes:
>
> Hi Stephen,
>
>> Retrieving newsgroup: rec.arts.sf.written...
>> Fetching headers for rec.arts.sf.written...
>> byte-code: End of file during parsing
>
> Before trying to enter that group, do M-x toggle-debug-on-error RET and
> post the backtrace here.

This is what I got:

Debug on Error enabled globally
Retrieving newsgroup: rec.arts.sf.written...
Fetching headers for rec.arts.sf.written...
byte-code: End of file during parsing
Commands: d, s, x, u; f, o, 1, 2, m, v; ~, %; q to quit; ? for help.

At the moment it looks like no clear message from the debugger,or it
is not entering it given the Commands line.

-- 
Stephen Harker                           s.harker@adfa.edu.au
PEMS                http://sjharker.customer.netspace.net.au/
UNSW@ADFA

Re: End of file during parsing, single newsgroup

[William Gardella]

Stephen Harker <sjharker-ooduxAEi7gXtt0EhB6fy4g@public.gmane.org> writes:

> Tassilo Horn <tsdh-mXXj517/zsQ@public.gmane.org> writes:
>
>> Stephen Harker <sjharker-ooduxAEi7gXtt0EhB6fy4g@public.gmane.org> writes:
>>
>> Hi Stephen,
>>
>>> Retrieving newsgroup: rec.arts.sf.written...
>>> Fetching headers for rec.arts.sf.written...
>>> byte-code: End of file during parsing
>>
>> Before trying to enter that group, do M-x toggle-debug-on-error RET and
>> post the backtrace here.
>
> This is what I got:
>
> Debug on Error enabled globally
> Retrieving newsgroup: rec.arts.sf.written...
> Fetching headers for rec.arts.sf.written...
> byte-code: End of file during parsing
> Commands: d, s, x, u; f, o, 1, 2, m, v; ~, %; q to quit; ? for help.
>
> At the moment it looks like no clear message from the debugger,or it
> is not entering it given the Commands line.

Stephen,

Could the "end of file during parsing" be corruption in the
~/.newsrc.eld file?  That's my only (possibly pretty far-fetched) guess.

-- 
I use grml (http://grml.org/)

Re: End of file during parsing, single newsgroup

[Tassilo Horn]

William Gardella <gardellawg@gmail.com>
writes:

>>> Before trying to enter that group, do M-x toggle-debug-on-error RET
>>> and post the backtrace here.
>>
>> This is what I got:
>>
>> Debug on Error enabled globally
>> Retrieving newsgroup: rec.arts.sf.written...
>> Fetching headers for rec.arts.sf.written...
>> byte-code: End of file during parsing
>> Commands: d, s, x, u; f, o, 1, 2, m, v; ~, %; q to quit; ? for help.
>>
>> At the moment it looks like no clear message from the debugger,or it
>> is not entering it given the Commands line.

Strange.  What does it say when you hit `?'?

> Could the "end of file during parsing" be corruption in the
> ~/.newsrc.eld file?  That's my only (possibly pretty far-fetched)
> guess.

I think, if it was corrupted it would error as soon as gnus tries to
read it the first time and not only with this one specific group.

Some more guesses:

  - Do you have (gnus-compile) in your ~/.gnus.el?  If so, comment it
    out and restart emacs.  Fire up Gnus, do something and hit `s' in
    *Group* so that it saves the ~/.newsrc.eld.  Restart emacs and gnus
    and try to enter rec.arts.sf.written.  I think, you still get an
    error but maybe a bit more informative.

  - Do you have a special gnus-summary-line-format set for that group,
    e.g., via gnus-parameters?  Maybe that's somehow broken?

  - Did you use a newer version of emacs once?  If so, maybe the
    (gnus-compile) compiled something in ~/.newsrc.eld to byte-code
    that's not understood by emacs 22.

Bye,
Tassilo

Re: End of file during parsing, single newsgroup

[Stephen Harker]

Tassilo Horn <tsdh@gnu.org> writes:

> William Gardella <gardellawg@gmail.com>
> writes:
>
>>>> Before trying to enter that group, do M-x toggle-debug-on-error RET
>>>> and post the backtrace here.
>>>
>>> This is what I got:
>>>
>>> Debug on Error enabled globally
>>> Retrieving newsgroup: rec.arts.sf.written...
>>> Fetching headers for rec.arts.sf.written...
>>> byte-code: End of file during parsing
>>> Commands: d, s, x, u; f, o, 1, 2, m, v; ~, %; q to quit; ? for help.
>>>
>>> At the moment it looks like no clear message from the debugger,or it
>>> is not entering it given the Commands line.
>
> Strange.  What does it say when you hit `?'?

That is taken from the message buffer.  I don't get the debugger
buffer unless I create it from another error.  If I hit '?" to TAB I
get the following:

SPC:Select  n:Forward  DEL:Backward  q:Exit  C-c TAB:Run Info  ?:This help [2 times]
indent-relative: Buffer is read-only: #<buffer *Group*>

>> Could the "end of file during parsing" be corruption in the
>> ~/.newsrc.eld file?  That's my only (possibly pretty far-fetched)
>> guess.
>
> I think, if it was corrupted it would error as soon as gnus tries to
> read it the first time and not only with this one specific group.

I could not find any obvious problem in the .newsrc.eld file.  I tried
copying a .newsrc.eld file from a backup prior to the error starting
and find the same error.

> Some more guesses:
>
>   - Do you have (gnus-compile) in your ~/.gnus.el?  If so, comment it
>     out and restart emacs.  Fire up Gnus, do something and hit `s' in
>     *Group* so that it saves the ~/.newsrc.eld.  Restart emacs and gnus
>     and try to enter rec.arts.sf.written.  I think, you still get an
>     error but maybe a bit more informative.

I don't have a ~/.gnus.el, all values set are done manually in my
~/.emacs, which has not changed in some months.  I am about to try
this suggestion (and some of the others).  

>   - Do you have a special gnus-summary-line-format set for that group,
>     e.g., via gnus-parameters?  Maybe that's somehow broken?

I don't have it set anywhere that I can find with grep.

>   - Did you use a newer version of emacs once?  If so, maybe the
>     (gnus-compile) compiled something in ~/.newsrc.eld to byte-code
>     that's not understood by emacs 22.

That is a possibility.  I started an install of Fedora 17 and don't
think I did anything in the home partition, but can't guaranteee
it.  

-- 
Stephen Harker                           s.harker@adfa.edu.au
PEMS                http://sjharker.customer.netspace.net.au/
UNSW@ADFA

Re: End of file during parsing, single newsgroup

[Stephen Harker]

Stephen Harker <sjharker@netspace.net.au> writes:

> Tassilo Horn <tsdh@gnu.org> writes:
>
>> William Gardella <gardellawg@gmail.com>
>> writes:
>>   - Did you use a newer version of emacs once?  If so, maybe the
>>     (gnus-compile) compiled something in ~/.newsrc.eld to byte-code
>>     that's not understood by emacs 22.
>
> That is a possibility.  I started an install of Fedora 17 and don't
> think I did anything in the home partition, but can't guaranteee
> it.  

This led to a solution.  I used Fedora 17 which comes with emacs 24, I
think it was emacs-24.1-4.  Running gnus with the later emacs and
opening rec.arts.sf.written resulted in an error message saying there
was a fault in the rec.arts.sf.written.ADAPT file, but proceeding and
overwriting it after exiting with a version that works with the older
emacs also. 

When I replaced the rec.arts.sf.written.ADAPT with a backup version I
must not have gone far enough back.  It may have been the first
version in which the error was written to the file.  I thought I had
gone back two days, but evidently not.

-- 
Stephen Harker                           s.harker@adfa.edu.au
PEMS                http://sjharker.customer.netspace.net.au/
UNSW@ADFA