Re: warning for index entry with :

[Patrice Dumas]

On Thu, Sep 29, 2011 at 11:55:28PM +0000, Karl Berry wrote:
>     My preference would be keep the warning as is until there is a way
>     to escape all the characters that cannot be in node names.
> 
> BTW, my thought on this has been to do something like this:
> 
> address@hidden address@hidden
> 
> I.e., same basic syntax as the image and index directives.  The
> arbitrary chars would use some escape syntax, maybe _XXXX as in html
> xrefs, to avoid the problematic characters (comma, colon, etc.) from
> occurring literally.

At first sight, that look ok.  One difficulty is that it should be done
so not only in node header, but also in menu entries (and @ref).

> In any case, we should only use the new convention in cases where the
> name was disallowed before, so that Info files keep working with older
> viewers as well as they do now.

Agreed.

> Oh, I guess there is one new complication, --plaintext output.
> Evidently we currently ignore @image with --plaintext (which seems
> potentially wrong, shouldn't we output the alt text if it's present?),

In tp, @image is not ignored, it is replaced by the basename.txt
content.  Alt could indeed be used, too.  I'll put that in the TODO
if I don't forget.

In fact, the Info converter inherits methods from the Plaintext
converter and redefine only the methods one that should be different, 
namely the main output method and _contents, _printindex, _node, _image.
Also for plaintext SHOW_MENU = 0.

> but we can't just ignore node names and the like.  

In fact we do in most cases.  Indeed, there are no node headers in 
plaintext, menus also are ignored, so the only place where node names occur 
are @*ref cross references.  It would be pretty straigthforward to have
a difference between info and plaintext on cross refs.

> Since for plain text
> there is no need for things to be machine-parsable, I guess we can just
> output the commas and colons as themselves?

Indeed.

> I doubt we want to do all this for the upcoming release, we're making
> enough major changes, already, but just getting it on record.  wdyt?

Agreed.

My Todo list is currently along

* trimming of customization doc and update of texinfo.txi
* tp/ modules documentation

(release?)

Then, in parallel
* FIXME in code and some TODO items in tp/
* new HTML customization API
 * porting of existing HTML init files to the new API
 * improvements of parts of the new API that are not correct
 * documentation of the HTML customization API
* pod2texi
* mysterious activity ;-)
* tex4ht and latex2html for @tex and @math

And after that, really new stuff, not in any particular order
* in-depth check of tests to get rid for real of texi2html tests
* code to transform Texinfo XML back to texinfo, for the translation stuff
  of Bruno -- if nobody else does it.
* LaTeX translator
* mediawiki translator
* new features (like the subject of this mail, but also new math commands, 
     @inline...)
* roff translator (if somebody is interested)
* CPAN module

-- 
Pat