[XeTeX] hyperref broken??

[Ross Moore]

Hi Will, and others,


On 04/10/2010, at 9:56 AM, Will Robertson <wspr81 at gmail.com> wrote:

> Am 03.10.2010 um 15:52 schrieb Will Robertson:
> 
>> On 03/10/2010, at 11:41 PM, Philipp Stephani <st_philipp at YAHOO.DE> wrote:
>> 
>>> Am 03.10.2010 um 14:02 schrieb David Perry:
>>> 
>>>> If I omit the option, or use 'xetex' in place of 'hypertex' or use the driverfallback option, I get:
>>>> 
>>>> ! Undefined control sequence.
>>>> \str_if_eq_p:nn #1#2->\tex_ifnum:D \pdf_strcmp:D
>>>>                                              {\etex_unexpanded:D {#1}}{\...
>>>> l.479 ...p:w  \cs_set_nopar:cpn  \cs_set_nopar:Npn
>>>> 
>>>> I fixed the missing commas but that made no difference.
>>> 
>>> Looks like an expl3 problem: \pdf_strcmp:D is pdfTeX's \pdfstrcmp, which might be nonexistent or available under a different name in XeTeX. This *might* get fixed by updating expl3; otherwise write to the LaTeX list (LATEX-L at LISTSERV.UNI-HEIDELBERG.DE, I've put it on CC).
>> 
>> I think I fixed this coincidentally earlier today. The update will roll out in a timely fashion.  

On the face of it, this looks similar to the problems I showed you yesterday.
Is that what you are saying your latest fixes address?

So the first question should be what is the version of XeTeX that MikTeX is providing.
And is David actually using that, or still an earlier version, as I was on a Mac.
If the primitive is not there, then the latest versions of expl3 and fontspec just will not work, and may not even succeed in loading successfully. Is that not correct?

> 
> Now that I sit down with the code in front of me, I can't comprehend where this problem is coming from.
> 
> Oh, maybe the problem is that some dastardly package is setting \XeTeXversion to \relax? But then, no, this would only cause problems in non-XeTeX use, from what I can see. I can't work this out without stepping through the actual problem.
> 
> Can someone post a minimal example that fails under MiKTeX? (Including \listfiles in the preamble.)

Yes; this is what we need to see; especially the first line of the .log, with the executable's version. 

> 
> -- Will


Cheers,

         Ross