[XeTeX] \XeTeXrevision (was: A XeTeX bug or a MiKTeX bug?)

[Jonathan Kew]

On 27 Apr 2008, at 6:31 pm, Andreas Matthias wrote:

> Jonathan Kew wrote:
>
>> [XeTeX] is updated to v. 0.998.1, the problem should go away.
>
> And other problems arise. With this new revision number the
> following check doesn't work any more:
>
> \ifnum\XeTeXversion=0
>   \edef\temp{\expandafter\@gobble\XeTeXrevision}%
>   \ifnum\temp<997
>     \@latex at warning{no way}%
>   \fi
> \fi
>
> Thus, a more complex test is needed:
>
> \def\gobble at nil#1\@nil{}
> \ifnum\XeTeXversion=0
>   \afterassignment\gobble at nil
>   \count@\expandafter\@gobble\XeTeXrevision.\@nil
>   \ifnum\count@<997
>     \@latex at warning{no way}%
>   \fi
> \fi
>
>
> Well, at least in my texmf-tree pdfpages seems to be the only
> package that uses \XeTeXrevision. I will fix it.

Oops. Sorry about that. Actually, it also caused trouble for a  
similar check in the config file "unicode-letters.tex"; I have just  
updated this in the xetex and texlive repositories.

(IIRC, a rev with the form ".997-dev" has been in circulation  
previously, so that could have caused problems too.)

>
> Will only development version have this kind of minor revision
> number (the number after the second dot)? How about having a
> XeTeX_revision_minor?
>
> @d XeTeX_version=0
> @d XeTeX_revision==".998"
> @d XeTeX_revision_minor==".0"
> @d XeTeX_version_string=='-0.998.0' {current \XeTeX\ version}

No, I think I should try to be more consistent about bumping this  
minor rev (or bug-fix) number, so that people can tell exactly what  
they've got. So I expect TeX Live 2008 to ship with XeTeX labelled as  
0.998.1, or a later "minor revision" if there are further bug-fixes  
before the final build there.

>
> Or putting the minor revision number only into XeTeX_version_string,
> the way pdftex does it?
>
> @d XeTeX_version=0
> @d XeTeX_revision==".998"
> @d XeTeX_version_string=='-0.998.0' {current \XeTeX\ version}

OK, that seems like a good idea. I'll consider that for 0.998.2,  
assuming we have another bug-fix sometime soon. :)

Meanwhile, making the test in pdfpages more tolerant of strange  
revision strings would be a good thing, given how I've changed them  
over time. My apologies for the hassle I've caused you!

JK