Re: @value an user macro expansions in ignored @inline*?

[Patrice Dumas]

On Tue, Sep 03, 2013 at 11:21:38PM +0000, Karl Berry wrote:
>     I can't see why ignored @inlineraw wouldn't have side effects, while
>     ignored @inlinefmt would have.
> 
> I certainly agree in principle.  Can you show me an example so I can
> check TeX?  My expectation is that the ignored text has to be
> syntactically valid (e.g., no unmatched braces) but doesn't have side
> effects.

Actually, I just tested with

\input texinfo.tex

@setfilename g.info

@node Top
@top top

@macro mymacro{}
aaa
@end macro

@inlinefmt{html,
@unmacro mymacro
@macro mymacro{}
ggg
@end macro
}

@inlinefmt{tex,
@unmacro mymacro
@macro mymacro{}
hhh
@end macro
}


@mymacro{}

but TeX doesn't like it:

./g.texi:24: Missing @endcsname inserted.
<to be read again> 
                   @let 
@recursivefalse ->@let 
                       @ifrecursive address@hidden 
@macro ->@recursivefalse 
                         @parsearg @macroxxx 
<argument> mymacro @macro 
                          mymacro{} hhh @end macro @fi 
@\unmacro #1->@if address@hidden ismacro.#1
                                     @endcsname @global @cslet
{#1}{macsave....
l.24 }
      

>     The second issue is about @value and user defined macro expansions.
>     Is it really right to have those expanded in ignored @inline*?  
> 
> No expansion sounds like the right thing to do.
> 
>     What does TeX do?
> 
> I think it is just thrown away, no expansion.
> Can you show me the example?

I con't really imagine an example that shows something for TeX, as TeX
does not allow fancy stuff that could help test whether value or macros
are expanded.  Here is an example for texi2any, but that is certainly
invalid for TeX:

@set closebrace }

@inlinefmt{html,
truc @value{closebrace}
machin

-- 
Pat