Re: [Texi2html-bug] new command-line switches for makeinfo

[Patrice Dumas]

On Sat, Jul 17, 2010 at 10:55:10PM +0000, Karl Berry wrote:
> 
> I certainly agree that the performance is not an issue, but there *is*
> another reason not to enable it by default: enabling it by default makes
> it seem like the feature is part of the language.  It isn't.  The
> problem is that any document that uses @-commands in node names is
> highly unlikely to work with texinfo.tex.  I wish that wasn't the case,
> but it is, and it's not going to be changed any time soon, if ever.

Maybe, but, currently, --commands-in-node-names doesn't really allow 
@-commands in node names. It just makes sure that @value (and maybe 
@macro) are rightly expanded -- if I haven't missed anything. 

> That is why I always thought of it as being an optional feature that
> people should have to enable explicitly, being aware of the deficiency
> with texinfo.tex.  Nothing to do with performance.

So, you mean that --commands-in-node-names should change meaning and
if not set no @-command should be allowed in node names? I am not sure
that it is so much needed. texi2dvi supports all the @-commands 
in node names and ref that I tested (see below). I didn't test
@value or user defined macros (which are considered as broken in the 
documentation, although it is worded as if they are broken in makeinfo). 
But for those it is always possibleto use -E such that makeinfo does 
the expansion.

texi2dvi doesn't acceept @verb in @ref, but doesn't accept it in 
@chapter either. (makeinfo in C or in perl have no trouble with @verb at all).

More precisely, I tested that texi2dvi accepts:
A @sc{sc} node @"i @"address@hidden @address@hidden @l{} @,{@'C} @exclamdown{}
a @strong{strong} ref with @address@hidden a i trema @"address@hidden a dotless 
i trema @"address@hidden @address@hidden and exclamdown @exclamdown{}

(and oddly, since texi2dvi accepts @, in node names, and makeinfo in C 
doesn't, it even perform better than makeinfo in C ;-).

I agree that there are still quite a bit of @-commands specified, for 
exmaple in 'HTML Cross-reference Command Expansion' that I haven't tested,
but they are less likely to be useful. Testing makeinfo in perl on them is
useful anyway, so I have added a test to be done to the TODO.

> In any case, regardless of the default, I think the option should stay
> in the --help message (and documentation); otherwise, people will think
> it is no longer supported.  But we will support it forever, so as not to
> break scripts/people that are used to it.

I agree that we need to keep it forever, but do we really need to keep it
in documentation and --help? If people see that it is no longer here,
they change their call to makeinfo which is better. If they don't notice
nothing is broken.

-- 
Pat