>  > Are you all familar with Graham Williams catalogue? in many ways, he
 > no unfortunately not yet, but it looks very much as what we are
you clearly didnt read your TeX Live documentation either this year or
last year....

 > that would be a very useful thing to have
it is ready *is*. it has over 550 entries

 > may i suggest not to use "latex3" that will confuse people a lot if
 > they come across such an entry without the background --- and thus
 > also not latex1 etc. not sure what would be best.
1latex, 2latex, 3latex etc

 > that's fine with me although i would prefer to keep it 2a and 2b
 > because my view is that distributions should try to keep also 2b. but
 > fair enough.
ok, if thats the way you interpret 3, lets go back to 2a/2b

 > but i would still want my "does not work with current latex class"
 > which would be 6 in your classification. otherwise either those
ok, i take the argument. there is a class of package with doesnt work,
but is a useful place to look for ideas

 > hmm, that does probably need some discussion of its own as i don't
 > think that for most stuff "best" is really a criteria. take again my
 > chemistry example. xymtex should be in 3 (imho) or 2b as i would
 > prefer but so should phchtex by Hans as both use a completely
 > different approach to the subject.
thats a hard one. is it a common situation or an exception? i was just
thinking of ways to avoid having 45 different classes implementing The
Way Letters Must Be Written To German DIN Standards (gordon bennett,
is there a DIN standard for email too?)

 >  > I fear that the `nyj', `kluwer' and `elsevier' classes must
 >  > remain in 3, as they cannot be called `must have' or `best of class',
 >  > and so they will not end up in the main distributions.....
 > and that is exactly why i'm against 2/3 but for 2a 2b as well behaving
 > package should preferably be in the main distribution.
hang on, that doesnt agree with what i am saying, i think. my
criterion for 2a/2b is that the package be widely useful. does
`kluwer' fall into that category? well, possibly. could be argued about.

 > after all we are not talking about that much space and anyway space is
 > becoming cheaper and other programs tools also come these days with a
 > bigger set.
in that case, what is the distinction between 2 and 3?

 > entries, eg tipa is font related but also a special application
 > (namely phonetic alphabet support)
sure, its multiple keywords. but lets have a constrained set of

 >  > \special restrictions')
 > good point, but why not call that restrictions=
yes. actually, i think this is valuable info which we dont currently have

 > that is a good start, i would agree. still it doesn't hurt to make
 > provisions for the non-latex stuff as well, since if a tester test
Williams already covers much else besides LaTeX

this boils down largely to a plea:

 a) use Williams catalogue and give him feedback
 b) supply info to him for extra keys in the database
 c) encourage distributions to use the catalogue