[XeTeX] revTeX + xltxtra + figure

[Arno Trautmann]

Hi Ross!

Ross Moore wrote:
> On 28/06/2009, at 5:29 PM, Arno Trautmann wrote:
> The incompatibility is with  revtex4  and  fixltx2e .
> Try the following start to your example, with the
> last line either commented or not. You should see
> the same error, since  xltxtra  requires  fixltx2e.sty .

Yes, indeed, the problem is with fixltx2e. I would have never came to
that idea …

>> ! Missing number, treated as zero.
>> <to be read again>
>>                    \aftergroup
>> l.35 \end{document}
>> ––
>> However, using fontspec and xunicode directly instead of loading xlxtra
>> works fine.
>> Can anybody give me a hint what I am doing wrong?
> 
> I don't know why  xltxtra.sty  has the line
> 
>   \RequirePackage{fixltx2e}[2006/03/24]
> 
> Surely Will can explain, when he gets a chance to do so.

I guess because it fixes some problems in the kernel. So in principle,
it should make sense to load it …

> You could try taking a local copy of  xltxtra.sty  and
> commenting-out this line.  Watch out for anything else
> breaking, and report it back here.

Without fixlxt2e, everything seems to be fine (at least regarding
figure[*]).

>> Fileversions:
>> ––
>> This is XeTeXk, Version 3.1415926-2.2-0.999.6 (Web2C 7.5.7)
>> xltxtra.sty    2008/07/29 v0.4 Improvements for the "XeLaTeX" format
>> ––
>> revTeX4 redefines figure in the following way, maybe this helps:
>> ––
>> \newenvironment{figure}
>>                {\@float{figure}}
>>                {\end at float}
>>
>> \def\@float#1{%
>>  \@ifnextchar[{%}]{%Brace-matching klootch
>>   \@yfloat\width at float{#1}%
>>  }{%
>>   \@ifxundefined at cs{fps@#1}{%
>>    \edef\reserved at a{\noexpand\@yfloat\noexpand\width at float{#1}[\csname
>> fps@\endcsname]}\reserved at a
>>   }{%
>>    \edef\reserved at a{\noexpand\@yfloat\noexpand\width at float{#1}[\csname
>> fps@#1\endcsname]}\reserved at a
>>   }%
>>  }%
>> }%
>>
>> \def\end at float{%
>>  \end@@float{%
>>   \check at currbox@count
>>  }%
>> }%
> 
> This kind of change could be relevant, as fixltx2e.sty
> seems to have a lot of coding to handle floats differently
> within a 2-column setup.
> 
> Tracing this is pretty difficult.
> I'd like to hear from some of the package writers,
> to better understand how their coding efforts have
> created some kind of clash.

That would be great! So far, I can get along commenting out fixltx2e,
but that cannot be a real solution.

Thank you very much for your explanations, that will help me very much!

cheers
Arno

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 260 bytes
Desc: OpenPGP digital signature
URL: <http://tug.org/pipermail/xetex/attachments/20090628/bced82ed/attachment.bin>