[tex-hyphen] shortdesd, longdesc of hyphen-<language> & CTAN

[Robin Fairbairns]

Mojca Miklavec <mojca.miklavec.lists at gmail.com> wrote:

> Dear Karl,
> 
> On Sun, Jun 5, 2011 at 01:09, Karl Berry wrote:
> >
> > Will there be an update of hyph-utf8 for TL'11?
> 
> Apart from what I wrote earlier, I plan to write "shortdesc" and
> "longdesc" into tlpsrc. Currently the strings are randomly taken from
> CTAN catalogue, often stating that "this package is deprecated" (as it
> really links to the old hyphenation patterns where CTAN team added
> that it is obsolete to use those old patterns).

i had understood that shortdesc and longdesc *always* came from ctan, if
ctan is the source.

> The majority of
> packages entirely lacks any description at all.

there are certainly ctan packages that lack a description, or even (in
some cases) a catalogue entry.  hyphenation patterns often lack
description, since the original authors often submitted them without any
explanatory text -- in which case the catalogue entry only contains
a description if i happen to know something about the background.

i can certainly plug in a description for one that's missing it, given
some text; if you'll alert me to where you've updated tlpsrc (and hence
texlive.tlpdb) i can back-fit it.  what i don't know is whether there is
a danger of things not appearing in tlpdb if they're not in the relevant
catalogue entry.

> Do you have any particular requests of what I should put there? I plan
> to create a short description
>     "Hyphenation patterns for English/Polish/Lao/..."
> and a long description containing:
> - author(s)

that information is best held in the catalogue, surely?

> - encoding for 8bit engine

useful to have.

> - (if something special has to be told, like in case of german that
> loads different patterns for 8bit and utf8 engines)
> 
> Anything else?
> 
> And once I do that ... I'm still thinking about the idea to
> auto-generate a CTAN-friendly version of hyph-utf8.

istm you're talking about a unknowledgeable-ctan-friendly version.  what
we have now seems fine, as far as making the stuff available to those
who know what they're doing.

> That is, a special
> folder hyphen-<language>, where each language would contain (symbolic
> links to):
> - hyph-foo.tex
> - loadhyph-foo.tex
> - conversion file used for utf8->8bit (not sure about that)
> - hyph-foo.{pat,hyp,lic,chr}.txt (not sure about that)
> - phyph-foo.tex (not sure about that)
> - some readme file explaining what is there, what needs to be edited
> in case of changes and with a link to the original file if needed
> - (and then we could also have a proper entry in CTAN catalogue
> pointing to the folder in question)
> 
> Any comments about that? (Hmmm, and I would probably need to extract
> the licence somehow?)
> In case that we go though all the files & languages and would want to
> contribute to entries in CTAN catalogue - what else is needed for
> that?

i create catalogue entries on the basis of the package that's been
submitted, mostly.  the catalogue dtd has provision for things that
aren't known (author "unknown", licence "noinfo", etc), but mostly i
just leave things out when the author hasn't supplied information.

hyphenation is possibly a more complicated case, since there's currently
so little information in most hyphenation files...

robin

(who still hasn't finished his cataloguing work from yesterday, hence
the delay in replying...)