Hello, about a year ago I announced the intention of letting go of Debtags[1]. No successor has happened to carry over maintenance since then, so it's time to follow my own advice[2] and properly let go. # What I tried, what worked, what didn't Debtags has existed for almost 20 years, with the intention of providing a categorization system for Debian packages, as a better way than sections to explore Debian's vast and growing body of packages. At the time I set off, studied some library science, found faceted classification as a useful theoretical foundation, and together with others slowly built a vocabulary of categories that are tailored to Debian's specific structure. I my past I used to try way too hard to be helpful, and ended up with an excessively overdesigned system. In later years I grew up wiser and worked to significantly simplify everything as much as I could (and I could probably simplify it even more). I implemented assisted ways of categorising packages, to lower the barrier of entry for people trying to categorise packages. I wrote guidelines to having a consistent tag vocabulary, and let any DD commit to the vocabulary itself so that it can be community-maintained. I created a way to anyone with a Salsa account to contribute to tagging and be acknowledged for it on contributors.debian.org I tried (and failed) to get an automatic route for categories to move from debtags.debian.org to ftp.debian.org, so that I am not in the critical path of people seeing their work appear in the Debian archive. # What I would have tried next I've been having a plan of mixing debian/control and debtags.debian.org as authoritative sources. The idea is that if a faced is used in debian/control, then debian/control is authoritative for that facet. The final list of tags in the Packages file will then be built by taking all tags in debian/control, and adding all tags from debtags.debian.org for all facets not listed in debian/control. # Accept the grief and let go I understand that the use case of exploring a distribution as a whole is nowadays mostly a minority use case, and people mostly find packages by searching the internet and then checking if what one found is packaged in Debian. As much as I would like to let go of Debtags in a way that it keeps existing without my involvement and other DDs can step in to maintain it, I cannot do it without significant involvement of other teams like ftp-master that have more pressing priorities. It's been a great ride, and I'm happy of what we all achieved. I think this is one of those aspects where Debian has managed to produce nontivial innovation. That innovation however over time hasn't gained enough traction, and it's time to let go. # What next From now on, as I have free moments, this is what I plan to do: 1. upload the latest dataset from debtags.debian.org to Debian 2. take debtags.debian.org offline 3. orphan/remove the debtags package Existing tags will remain in the ftp-master overrides file until ftp-master will decide to remove them. Since the Tag: header is currently optional in package data, applications should continue working fine when the data will at some point disappear. I will make an effort to try to preserve sources and data for everything somewhere purely for historical purposes. [1] https://lists.debian.org/debian-devel/2022/10/msg00248.html [2] https://www.enricozini.org/blog/2023/debian/adulting/slides.pdf#page=18 [3] https://wiki.debian.org/Debtags Thank you all for this fantastic ride! Enrico -- GPG key: 4096R/634F4BD1E7AD5568 2009-05-08 Enrico Zini <enrico@enricozini.org>
Attachment:
signature.asc
Description: PGP signature