Are you all familar with Graham Williams catalogue? in many ways, he has already laid the foundation for what we are talking about, with a BiBTeX database looking like this: @TeXIndex{akletter, texlive = {latex3}, ctan = {macros/latex/contrib/supported/akletter/}, abstract = {An advanced letter document class which extends \LaTeX{}'s usual letter class. Provides support for building your own letterhead; can mark fold points for window envelopes; and more. Documentation in German, but sample file is sufficient for a basic understanding.}, author = {Axel Kielhorn}, email = [log in to unmask], modified = {26 Mar 1997 09:55:33 [log in to unmask] } the `texlive' key (renamed) can be used for what we seem to agree is the useful scheme: latex1: a core tool provided by Them (eg graphics); distributions without all of these are incomplete latex2: a generally agreed important tool package (eg calc) latex3: a generally agreed important application package (eg tipa) latex4: a package that works under the current LaTeX, has no known bad behaviour, but is not a `must have' (eg nassflow) latex5: a package of unknown status, may or may not work, perhaps whose author is not contactable any more I use 2 and 3 instead of 2a and 2b, because I think we *do* mean that distributors should treat this as a priority list when packaging up. the goal would be to persuade people to take *all* of 2 and 3, or just all of 2, and not make their own arbitrary decisions. Entry to 5 is the default; entry to 4 is (in my book) when it can pass a trivial CTAN to TDS test, ie it processes itself and its documentation, and fits into the TDS tree. That does not require any decisions, just a volunteer testing it, and reporting to the coordinator. Entry to 2 or 3 is by acclamation and discussion. Inevitably, that means Yet Another Mailing List, I think. Sigh. But not impossible. That group *must* be strict - we want to keep the number of packages in 2 and 3 relatively small. The `best of class' rule must apply - if there is already a float-extension package there, a new one will not be admitted unless the old one is thrown out. 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..... Then a new key can be added, `class', to select from the following list: - extensions to normal facilities (tables, floats, cites etc) (eg mdwtab) - class files for publishers and journals (eg kluwer) - class files for theses, resumes etc (eg suthesis) - letters (eg akletter) - font related (eg tipa) - language related (eg french) - graphics (eg xypic) - packages for special applications: Chemistry, Computer Science, Mathematics, Physics etc. (eg xymtex) - hypertext-related (eg hyperref) - tools (eg calc) and possibly that `compatibility' key (including, `does it have \special restrictions') SO, where does that leave us in actions? 1. I suggest Graham Williams be asked to act as repository of information in his catalogue, if he agrees to have 2 keys supplied by other people 2. We need one or more coordinators to enter new packages in at level 5 and then 4 3. We need a forum to allow entry to 2 and 3 4. We need to publicize the scheme and ask packagers to adopt it. Frank et al can help with moral pressure (in my book the key is that `put in all of 2 or none of 2' I am happy to volunteer as a coordinator, and I can reasonably promise (since I am still Secretary) the services of the TUG computer for mailing list, Web site and repository. If this takes off, it fulfils one of my aims for TUG and UKTUG (at least), which is to provide a more stable basis for the TeX Live CD ROM distribution. It doesn't classify the whole CD, but it covers a good portion of. NOTE that I am not proposing a full CTAN classification! Just those things which are plainly `LaTeX contributions' Comments? Sebastian