[XeTeX] E not lowered in \TeX in .995

[Bruno Voisin]

Le 25 août 06 à 02:05, Jon Breitenbucher a écrit :

>> On Aug 24, 2006, at 6:45 PM, Bruno Voisin wrote:
>>
>>> In order to help somebody isolate the source of your problem, you
>>> should try removing input and package calls one by one in your job
>>> which doesn't work, until the problem disappears. Both runs seem to
>>> use the same version of xltxtra.sty. The problem must be elsewhere:
>>> incompatible packages, probably.
>
> Yes you are correct. But the changes in the code from v.1 to v.2 seem
> minimal and I can see NO reason why there should be a conflict in v.2
> and not v.1. We are setting some lengths and then feeding them values
> in v.2 where as in v.1 we don't set some lengths but feed the values
> in directly. Am I misunderstanding the changes to these particular
> commands? For now the easiest thing for me is to use the old code
> that works.

If we are able to isolate which specific package call in username.tex  
(your job which doesn't work) makes \XeTeX faulty in version 0.2 of  
xltxtra.sty and correct in version 0.1, then we'll be able to correct  
the bug in this version 0.2.

So we really need a minimal example, a reduction of the "job which  
doesn't work". It's practically impossible to guess from your very  
long list of included files which package may be a likely culprit.  
Possibly setspace.sty. Or this means somebody has to get copies of  
all the packages that your job is calling, and then study the code of  
each of them in detail to see which part may affect the working of  
xltxtra.sty v2. That won't be me!

One possible explanation might be that xltxtra.sty uses the "macho"  
TeX programming style that the LaTeX companion warns against: i.e.  
using plain-TeX control sequences such as \hbox, \lower and \raise  
instead of the corresponding standard LaTeX commands \mbox and  
\raisebox; or using \newlength and \setlength without putting their  
arguments inside braces { and }. While normally harmless, this might  
cause conflicts with other packages playing with the inner TeX  
workings; whereas -- in an ideal LaTeX world -- redefinitions  
adhering closely to the LaTeX syntax rules should be safe.

That's also what the LaTeX 3 team is warning against, for a different  
reason: while the low-level commands used internally by LaTeX (in its  
classes and packages) may work differently or not at all in LaTeX 3  
vs. LaTeX 2e, the top-level commands (i.e. those documented and  
directly accessible to the user) should be backwards-compatible.

But without a minimal example, this is just useless speculation.

Bruno