[tex-live] Catalogue sources

[Robin Fairbairns]

On Fri, Sep 11, 2009 at 11:31 PM, Karl Berry <karl at freefriends.org> wrote:

>    >   catalogue-date 2007-02-24 15:09:57 +0100 (Sat, 24 Feb 2007) $
>    >
>    > All other packages don't have anything after timezone.
>
>     We take the date string 1-1 from the TeX Catalogue, so it should be
> fixed
>    there. Robin?
>
> It's not Robin's problem.  I see nothing special about the datestamp
> item in parrun.xml vs., say, paresse.xml.   I expect our parsing is
> somehow failing, or it's an old thing which hasn't gotten updated.  I'll
> look into.
>

we don't _write_ those dates -- they're from svn, inserted at the point
of checkin.  we can't therefore "correct" them.  if someone wants them
in some sense "right" in tlpdb, they should undertake the editing task
themselves: we changed to this method because we found ourselves
less than reliable putting the dates in by hand.


>    > * Several packages are not on ctan and therefore have empty
>    > catalogue-ctan:
>    >   colorsep
>    >   dvips
>

we do have a dvips directory.  it contains a readme saying "look
at tex live.

   >   gustlib
>    >   kpathsea
>

seems to be a part of the metapost distribution.  should i catalogue
it as coming from there?  (ha ha,  that's a _joke_.)


>    >   latex
>

we do actually hold that(!).  i guess i'll put a path entry into the
catalogue; any such entry is misleading, one way or another, which
is why i've never done it before, but given the overwhelming
demand for some abstract property i'm not sure about, i'll do it some
time.


>    >   texdoc
>    >   times
>

looking at this entry, i would guess this is some attempt to fend off
complaints by people who can't understand the tool that times is
a small part of.  of course we hold it, but should the catalogue for
times hold a path entry pointing to the same place as the entry for
psnfss?  i judged not, i guess -- but it's been a long time.  should
we have similar entries for palatino, ..., zapf dingbats?

   >   xmlplay
>

all the otherwise unremarked entries are (afaics) distribution-only
packages.  this puts the shareware distributors at a bit of a disadvantage
(not that there are many of them, any more, so perhaps it's not a big
deal).

   > More curious than erroneous.

    That must be fixed in the Catalogue by adding
>            <ctan path='....'/>
>    I am sure that in one way or the other these things *are* on CTAN.
>

well, of course they're on ctan, but would a catalogue entry saying

  <ctan path='systems/win32/miktex/tm/packages/xmlplay.cab'/>

help anyone?

  Norbert, it's not so simple.  Many of them are actually not on CTAN,
  even though there may be Catalogue entries.


> And I don't see what <ctan path=> has to do with it.  I didn't think we
> depended on the path to find the descriptions.  Don't see why we should,
> at least.
>

robin

[i seem to have deleted the original of this message, hence replying from
my gmail account.  sorry about that.  i do have other ways of doing this,
but gmail (apart from the ghastliness of using web mail) is easiest.]
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://tug.org/pipermail/tex-live/attachments/20090912/71a23c13/attachment-0001.html>