[XeTeX] [lltx] unicode-math breakage with newest TL '10 update

[ulrik.vieth at arcor.de]

I've traced it with logging all and it seems to be an expansion
or catcode problem:

It happens during the processing of \um_input_math_symbol_table
for \g_um_mathtable_tl ...n {"0220F}{\prod }{\mathop }

It ends up trying to do a \global \def ∏, which would only work
if  ∏ is an active character or if ∏ were an unexpanded \prod.

Really messy stuff to debug.
Regards, Ulrik


! Missing control sequence inserted.
<inserted text> 
                \inaccessible 
<to be read again> 
                   &#8719;
<argument> \group_end: {\global \def &#8719;
                                        {\csname \cs_to_str:N \prod _sym\end...

\::: #1->#1
           
\um_set_big_operator:nnn ...r:N #2_sym\endcsname }
                                                  \group_end: \um_set_mathch...

\g_um_mathtable_tl ...n {"0220F}{\prod }{\mathop }
                                                  \_um_sym:nnn {"02210}{\cop...

\\setmathfont ...lt }\um_input_math_symbol_table: 
                                                  \um_remap_symbols: \um_set...



----- Original Nachricht ----
Von:     Khaled Hosny <khaledhosny at eglug.org>
An:      ulrik.vieth at arcor.de
Datum:   14.09.2010 21:34
Betreff: Re: [lltx] unicode-math breakage with newest TL '10 update

> On Tue, Sep 14, 2010 at 09:28:51PM +0200, ulrik.vieth at arcor.de wrote:
> > I've tried manually undoing the change in luaotfload, but it didn't help,
> > so I'm no longer sure about my attribution of the problem to luaotfload.
> > 
> > It could equally well be caused by some other recent updates in TL,
> > such as fontspec, unicode-math or expl3.  Will, could you check?
> 
> I have hard time understanding Joel message (I still don't), but this
> don't look like a luaotfload thing (but who am I to give statements
> about errors involving TeX macros).
> 
> Regards,
>  Khaled
> 
> -- 
>  Khaled Hosny
>  Arabic localiser and member of Arabeyes.org team
>  Free font developer
>