Print

Print


Sebastian Rahtz writes:
 >  > i think this goes a bit out of hand and in different directions:
 >  > originally the suggestion was to collect class files that are around,
 >  > now we are talking about classifying the whole LaTeX universe.
 >
 > its sometimes not easy to distinguish pure class files from other
 > paraphernalia?

true there are grey areas but in most cases it is trivial enough
(something ending in .cls .... :-)

 >  > collecting class files means not only looking at CTAN but actively
 >  > asking publishers (somehow) to give their class files out if they have
 >  > any. many publishers do have class files that they send to people on
 >  > request.
 > others, like us, actively do not distribute their real working class
 > files...

too bad in which cases yours will not be on CTAN and will not get
there. but the starting point was that people would like to have more
cls files at their hands and that there are more out there then most
people know and or have access to.

 >  >     * 5 does definitely not work with current LaTeX (after checking)
 >
 > in that case it would not be kept on CTAN! i dont see the point

well, what a good point. only problem is that many such things
are. when i started my private survey i even found document styles for
version 2.08 on CTAN and i bet they are still there.

there are reasons for keeping at least some of those in 5 nevertheless

  a) might be 2.09
  b) we might keep the work as eventually somebody might upgrade them
  c) historians might welcome it :-)

what i wanted 5 for is stuff that fails with 2e --- no further check

once all the available material is classified in that way the CTAN
people could still decide to throw those out on into a
veryobsoletestuff directory or whatever. point is unless somebody goes
through all the stuff that is out there on CTAN we will never catch
that stuff


 >  >     * by default everything goes to 4 (which mean uncheck, might not
 >  >       work with current LaTeX)
 > you are approaching this from a 3rd direction, which is whether things
 > *work*.

so did you see below. and without that test at least in its basic form
i think the whole exercise is pointless

 >  >       an active process, eg one way as i see this could happen is a
 >  >       couple of volunteers with a coordinator is taking the current
 >  >       ctan dirs in chunks and check the packages/class styles, write
 >  >       one para for each that seems to work and sends this finding to
 >  >       the coordinator.
 > thats a hell of a lot of work, and not particularly robust or even
 > useful. what I want to know is will package X work with my package Y,
 > and the volunteer tester is not going to find that out for me.

no that can't be easily achieved i agree. but the minimal check is
already a big plus (the one you said you used for tex-live) and that
and nothing more i was suggesting

 > my test for TeX Live is whether it runs its own documentation and test
 > files. i threw out half a dozen packages on this criterion - if they
 > don't work internally i am not interested in them at all. this also
 > provides the test `does it run against current release, as opposed to
 > release when they wrote it'

but this is exactly what i suggested, everything starts out at *4 (ie
unchecked) then

  - test if it works on itself (ie documentation runs through, example
    comes out ...)

  - test if it works with current release

if either of those fails then it goes into *5 otherwise it goes into
*3 plus a line to the coordinator what it does (or a form filled out
that has been agreed on containing some classification info as far as
it can be determined easily)

of course most important is that it works on its own, for example
xymtex (chemistry) does really work very nicely (see Graphics
companion :-) but its documentation needs 209 compatibility code. that
would be good enough to go to *3 although i (if i would be the tester)
would suggest to the author to change \documentstyle ->
\documentclass.

in fact that one should go in my opinion to 2-applications (after the
process of acclamation that would need to be established)

and even if it would balk at \footheight unknown or something like
this in the documentation (but only there) i would (as a tester) put
it into *3 but having in that form a line saying, documentation a bit
shaky press return on ...

 >  >       with the suggestion that 2-general should be always included
 >  >       into a distribution but 2-applications as well if there is a bit
 >  >       of space left.
 > thats a fair distinction

woah, at least one thing finding your approval :-)

frank