Hi everyone, On 02.11.20 21:23, Graham Inggs wrote: > We are doing a roll call for porters of all release architectures. If > you are an active porter behind one of the release architectures [1] > for the entire lifetime of Debian Bullseye (est. end of 2024), please > respond with a signed email containing the following before Friday, > November 27: [...] This roll call will be especially important for Debian s390x, which otherwise is at risk to be dropped. There is currently no active porter within Debian. I still feel a little bit of responsibility but I'm unwilling to commit to another cycle - I already did not actively commit to the last one. The s390x port is unique in quite a few ways: 1) It is the last remaining big-endian port among the release ports. It is also the only 64-bit big-endian architecture that became an officially released port. 2) All installations run headless. The initial installation is done via console emulation. Such installations are also very unlikely to be used as remote desktop solutions, so some desktop packages (e.g. Javascript interpreters as part of browsers which then end up as the core of desktop environments) are effectively unsupported unless implemented in a fully portable fashion. 3) It is hard for volunteers to get access to mainframes beyond the porterbox. If someone wanted to run the port on a real machine, procuring hardware of an age that is still supported by Debian and getting it to run in a DC are both incredibly challenging. While we have in the past been able to arrange suitable access for 3) there is still the question if there is a real user base using the product. Ubuntu receives money from a deal with IBM to offer their distribution and thus they are committed to it. But they also completely forked their installation and architecture support packages away from what Debian has. There's probably some marginal value here from Ubuntu, given that doko@ maintains the toolchain in both Debian and Ubuntu - but there's absolutely no QA on the Debian side right now and I'd be totally unsurprised if the end result were broken. (As evidenced by the broken installer in the latest stable release.) 1) and 2) are the real challenges from a practical point of view: With no real continuous integration (although elbrus@ trying to improve on this is much appreciated), you end up with bit rot killing installations, upgrades, and really just packages people would try to use that might crash at runtime. The dependency pickle of s390x not supporting desktop packages and thus it needing to be special-cased in packages or entire package trees be removed from the architecture, just makes it worse. You don't need to be a DD or DM to step up, although it helps. Aurelien mentioned that he'd happily sponsor fixes if necessary. Kind regards and thanks for listening Philipp Kern
Attachment:
signature.asc
Description: OpenPGP digital signature