On Wed, Apr 22, 2020 at 1:11 PM Branko Čibej <br...@apache.org> wrote:
> On 22.04.2020 13:06, Nathan Hartman wrote: > > On Wed, Apr 22, 2020 at 3:13 AM Stefan Sperling <s...@elego.de> wrote: > >> The current prospected release date for 1.14 is May 6, which is no longer >> viable. The official soak period (which starts today) does not allow for >> a release this soon. We will also need to allocate at least another week >> to collect new signatures for the final release artifacts once they have >> become available. >> >> So if we stick to the official plan, provided we don't discover any major >> problems during the soak period, we can prepare 1.14.0 on the 20th of May >> and probably release on the 27th of May. >> >> Is this OK or would anyone prefer to expedite this process (e.g. by >> shortening the soak period)? > > > This being a LTS .0 release, I think we should stick to the official soak > as documented, unless someone has a really compelling reason to do > otherwise. > > But perhaps we could try to save the extra week (May 20-27) by parallel > processing: If no showstoppers are found by May 13th, we could > optimistically begin testing / signing of the rebranded release artifacts; > if on May 20th no showstoppers were found in -rc2 and the required > signatures are collected, then we could release immediately. My only > concern with this "parallel processing" approach is the risk of confusion > that could arise if we did discover some major issue in the final week. > > > Agreed. > I concur. No compelling reason to shorten the soak. Also, if we would want to discuss / change the rules regarding the soak, we should do so after 1.14.0 has been released, not during. I agree that a week of testing is probably not needed. For past .0 releases most of us that had tested / signed the RC simply signed off quite quickly with a simple note like "Verified that 1.X.0 is identical to 1.X.0-rcY modulo version number and harmless changes (docs, ... whatever); and I have tested an signed 1.X.0-rcY". > Also, I think we should consider getting the Python 3 fixes for Windows > into the release. They're only related to bindings and tests, so I don't > foresee any showstoppers. > Yes, I think it would be nice if we could state "complete Py3 compatibility", including the testsuite on Windows. I think the only problems are with the testsuite BTW (and who on earth even runs the testsuite on Windows ;-) ...). Oh, and also with "gen-make.py" on Windows when using "Debug" configuration (still have to bring that one to the list, sorry). I don't know the exact rules, but does backporting testsuite fixes cause a restart of the soak? If we can backport those and include them into the final 1.14.0 without restarting the soak, that would be ideal. Otherwise ... > On the other hand, waiting for .1 for that wouldn't be terrible. > ... yes, that's fine too. -- Johan
