Re: CVS Texinfo fails to build the Autoconf manual (from git)

[Patrice Dumas]

On Wed, Jul 18, 2012 at 09:25:41PM +0300, Eli Zaretskii wrote:
> 
> Macros never worked well with TeX, so -E was a band-aid that not
> always worked.

I don't understand.  If macro do not work well with TeX, -E is a way to
avoid expanding macros in TeX.  To me, it may be the way to go.

> > > also a problematic concept, because the choice of things it expands as
> > > opposed to those it leaves alone is such that cannot possibly work in
> > > some use cases.  Here, "work" means that passing the results of -E to
> > > makeinfo should give you the same output as running makeinfo once
> > > without the -E.
> > 
> > Do you have an example?  I think I get the general idea, but I can't
> > find an example.  Maybe this has to do with @if*?
> 
> Yes, @if* are the problem.

And you think it is not fixed by calling makeinfo with --iftex
--no-ifinfo?

> If that can be fixed, it would be a good thing.  OTOH, if the new
> macro expansion machinery will work well with TeX, perhaps -E could be
> deprecated altogether.

I don't really get what is the 'new macro expansion machinery'.
Basically, the new expansion machinery is the makeinfo in C expansion
without the inconsistencies and bugs.  So the incompatibilities with TeX
are still there, especially since these incompatibilities come, in
general, from the fact that we are limited in TeX.

Other changes, for example the fact that TeX constructs, such as 
@titlepage, are better taken in account in texi2any give more
incentives, if anything, to use -E.  I'd say that the only places where
TeX could lead to something different is in @everyheading and similar
and if pure TeX is used.

-- 
Pat