Hi Maxime, First of all, many thanks for your interest! Andreas Tille, on 2021-12-16: > Am Wed, Dec 15, 2021 at 11:19:27PM +0100 schrieb Maxime Chambonnet: > > Blame it on my poor knowledge of Debian, and the siloed internal search > > engine... I started today looking seriously into packaging ROCm, and after 8 > > hours discovered that there is a dedicated team, slightly hidden under the > > debian-ai mailing list : https://salsa.debian.org/rocm-team . > > Thanks a lot for fighting through sparsely documented situation and > finally making your way here. Yeah, the rocm-team area predates the construction of the debian-ai mailing list, hence the fragmented groups on salsa. > > The team have at least pushed one component of ROCm into stable already : > > https://packages.debian.org/source/stable/roct-thunk-interface. (old > > version, 3.7, current is 4.5, 5 announced) > > > > A conversation between an AMD engineer and the Debian ROCm team started here > > : https://lists.debian.org/debian-ai/2021/03/msg00009.html > > Seems you have checked the list archive and probably found ROCm relevant > threads. For instance there was some video call and if 'm not missleaded > some log of this (I neither took part nor have read the log). I feel I have some responsibility in the somewhat stalled discussions on the ROCm and must apologize for this; the month has been particularly busy so far on my end. Checking what I can do keep things to move forwards… > > > Would it be useful to place these packages under the umbrella of > > > Maintainer: Debian OpenCL Maintainers > > > <pkg-opencl-devel@lists.alioth.debian.org> > > > There is more in ROCm than just OpenCL, but I don't know a better > > > fitting team ... > > > > I have no clue how this could be better setup, between teams that deal with > > very similar issues. From quick lookup in rocm-cmake debian/control, the Maintainer was set to debian-ai@lists.debian.org, so I assume this would be the way to go? (There might be objections that it might create noise on the mailing list due to bug entries though, but let's consider moving ahead this way for the time being; surely it is possible to change that later.) > > There was no RFP nor ITP for the overall rocm, there is one now. > > Guess I will play catch up, and see if I can help! > > Beeing an outsider of ROCm topic I really welcome your contribution > anyway. So do I! Thank you for the nudge! Have a nice day, :) -- Étienne Mollier <emollier@emlwks999.eu> Fingerprint: 8f91 b227 c7d6 f2b1 948c 8236 793c f67e 8f0d 11da Sent from /dev/pts/2, please excuse my verbosity.
Attachment:
signature.asc
Description: PGP signature