[texhax] best way to revise a large existing text

W.J. Metzger wes at hef.ru.nl
Wed Oct 28 12:28:44 CET 2009

Dear Frederik,

This is what I get too for  latexdiff pap7.tex pap7.tex > pap7-diff.tex
But did you try it removing line 635?  That is the line
%%%%%%%%    }           ADDING THIS LINE PREVENTS segmentation fault in latexdiff
If I remove that line I get the segmentation fault.

Cheers, Wes

On Tue, 27 Oct 2009, Frederik Tilmann wrote:

> Dear Wes,
>
> I can't reproduce the error:  see the following transcript.
>
>> ~/tmp 56> latexdiff pap7.tex pap7.tex > pap7-diff.tex
>>
>> WARNING: Inconsistency in length of input string and parsed string:
>>      This often indicates faulty or non-standard latex code.
>>      In many cases you can ignore this and the following warning messages.
>>  Note that character numbers in the following are counted beginning after
>> \begin{document} and are only approximate.DEBUG Original length 109535
>> Parsed length 109533
>>
>>  in terms of $Q$ the \taumodel\ provides a good description, much bette
>>                               ^^^^^^^^^^^
>> Missing characters near word 7149 character index: 101911-101922 Length: 9
>> Match: |provides | (expected match marked above).
>>
>> WARNING: Inconsistency in length of input string and parsed string:
>>      This often indicates faulty or non-standard latex code.
>>      In many cases you can ignore this and the following warning messages.
>>  Note that character numbers in the following are counted beginning after
>> \begin{document} and are only approximate.DEBUG Original length 109535
>> Parsed length 109533
>>
>>  in terms of $Q$ the \taumodel\ provides a good description, much bette
>>                               ^^^^^^^^^^^
>> Missing characters near word 7149 character index: 101911-101922 Length: 9
>> Match: |provides | (expected match marked above).
>> ~/tmp 57> latexdiff --version
>> This is LATEXDIFF 0.5  (Algorithm::Diff 1.15 so)
>>   (c) 2004-2007 F J Tilmann
>> ~/tmp 58> perl --version
>>
>> This is perl, v5.10.0 built for i386-linux-thread-multi
>>
>>
>> Perl may be copied only under the terms of either the Artistic License or
>> the
>> GNU General Public License, which may be found in the Perl 5 source kit.
>>
>> Complete documentation for Perl, including FAQ lists, should be found on
>> this system using "man perl" or "perldoc perl".  If you have access to the
>>
>
> pap7-diff.tex seems to contain reasonable output.  Only change to pap7.tex is
> that some newlines get removed. (particularly before comments, or where there
> are multiple newlines)
>
> It is not the perl version either; I ran the same sequence on another
> machine, which has perlv5.8.0, and get the same output as above.  Also get
> the same result with latexdiff-so 0.5, and latexdiff-fast 0.42.
>
> If anyone else is reading this thread, can someone else reproduce?
>
> Your other reported bug (ignores "\ " is a real shortcoming leading to the
> warnings and  I will try to address this in the next version).
>
> Frederik
>
>
>
>
>
>
> On 27/10/09 16:20, W.J. Metzger wrote:
>> On Mon, 26 Oct 2009, Frederik Tilmann wrote:
>>
>>> Dear Wes
>>>
>>> I have never had any reports of segfaults, and I know some people have
>>> used
>>> it on their PhD thesis, so length should not really be an issue. It
>>> should
>>> really bail with a Perl error if there was anything wrong with the
>>> latexdiff
>>> code.
>>> What's your system and perl version? Did you try latexdiff-fast, which
>>> might
>>> be more robust if there is a memory problem with perl?
>>>
>>> Frederik
>>
>> Dear Frederik,
>>
>> I run on Scientific Linux 5.3, which is a clone of Red Hat Enterprise 5.
>> The perl version is v5.8.8 built for i386-linux-thread-multi
>> latexdiff latexdiff-fast and latexdiff-so all gave the segmentation fault.
>>
>> I tried doing it also on another machine with a slightly older version of
>> perl v5.8.5, but with twice the memory. It also gave the segmentation
>> fault.
>>
>> I've played around with the tex file and found that the segmentation fault
>> could be avoided by adding a comment line -- line 635 of the attached file.
>> If that line is removed, I get the segmentation fault.
>>
>> The segmentation fault occurs very quickly, almost immediately. So I think
>> that latexdiff has not started looking for the differences yet.
>>
>> I thought that the problem might be misinterpreting a { that was in a
>> comment, since adding a comment with a } got rid of the segmentation fault.
>> I attempted to isolate the problem in a small test file, containing only
>> the \begin{figure} - \end{figure} in which line 635 occurs. But I did not
>> get a segmentation fault with or without line 635.
>> So the problem is more complicated than just the { in a comment.
>>
>>
>> Another problem, but only a slightly annoying one, is an apparent
>> misparsing of a line ending in a \
>> e.g.
>> use \ell\
>> rather than l to avoid confusion with 1
>> Apparently the blank after \ell\ is not seen and results in warning
>> messages.
>>
>> Further, differences in equations sometimes lead to incorrect mathmode in
>> the difference file resulting in latex needing to insert a \$.
>>
>> All in all, latexdiff seems to work well for text, but has some problems
>> when things get complicated.
>>
>>
>>> W.J. Metzger wrote:
>>>> On Fri, 23 Oct 2009, martin f. krafft wrote:
>>>>
>>>>> also sprach Boris Veytsman <borisv at lk.net> [2009.09.22.1700 +0200]:
>>>>>> Try latexdiff, http://www.ctan.org/tex-archive/support/latexdiff/
>>>>>
>>>>> That was a marvelous suggestion. Thanks.
>>>>
>>>> It sounded good to me too. So I downloaded it and tried it -- works fine
>>>> on small tex files, but when I tried it on 'real' files it results in a
>>>> segmentation fault. Do others also have this experience?
>>
>> Cheers, Wes
>> --
>>
>> Dr. W. J. Metzger Experimental High Energy Physics Group
>> tel. +31-24-3653127 Faculty of Science
>> +31-24-3652099 (secr.) Radboud University Nijmegen
>> fax. +31-24-3652191 Heyendaalseweg 135
>> 6525 AJ Nijmegen, The Netherlands
>> e-mail: wes at hef.ru.nl or Wesley.Metzger at cern.ch
>> http://home.cern.ch/metzger/ or http://www.hef.ru.nl/~wes
>
>
> --
> Frederik Tilmann
> Bullard Laboratories           Tel. +44 1223 765545
> Department of Earth Sciences   Fax. +44 1223 360779
> University of Cambridge        email: tilmann at esc.cam.ac.uk