[tex-live] TL 2009: texdoc -l and ChangeLog and some others

[Manuel Pégourié-Gonnard]

On 06/01/2010 10:01, Robin Fairbairns wrote:
> and mostly (when they come to our notice) we rename them to just README
> -- which is what we ask for on upload.  i've renamed the eurosans one.
>
Thanks.

> however, we can't always be quite so cavalier.  for example:
> 
> 2000/05/16 |      38950 | fonts/psfonts/polish/antp/README.ENG
> 2000/05/16 |      39227 | fonts/psfonts/polish/antp/README.POL
> 
> obviously can't both be renamed; there are other packages where there
> are multiple READMEs that can't be treated simplistically like
> README.eurosans
> 
Sure.

> would it be possible for texdoc to grok the documentation entries in the
> catalogue?  i could process the catalogue into pretty much anything
> manuel would like, if it is of any use.
> 
Yes, there's already code for this in the development version of texdoc. More
precisely, texdoc gets the info from the texlive database, which in turn gets
info from the catalogue. (The point is that the directory components on CTAN,
hence in the catalogue entries, doesn't always match those in TL: even the
package name can differ.) So, the good news is, you don't need to worry about
processing it into a format.

Currently the idea is as follows: get all the entries in "docfiles" from the
database. It's rather crude. The next step is to use the metadata. The bad news
is, the content of the details attribute is free text, which is quite hard to
analyse the right way in texdoc. For example, "FAQ formatted for letter paper" :-)

I wonder if it would be feasable to add an attribute like "type" which would
take its values in a finite, well-determined set like
- usr : usr manual
- dev : commented source etc
- readme : readme files and the like
- ex : example document
- test : test document
- main : the main document(s), for documentation-only packages
- whatever proves to be useful.

While we're at it, it would also be useful if the language info was always
separated in its own attribute and also normalized.

I'm aware it probably means a lot of work. Actually, I've been planning to talk
with you (the catalogue team) about it for quite some time, but I wanted to
investigate more closely, and maybe propose a helper script, etc. But since you
raised the question...

Anyway, if you think the idea is interesting and need help in order to do these
quite extensive changes, I'd be glad to help (either with ideas, code, patches,
whatever is most convenient for you).

Manuel.