On Tue, Jun 4, 2024, 22:10 Eli Schwartz <[email protected]> wrote:
> On 6/4/24 3:37 PM, Joost Roeleveld wrote: > > > >> On 6/4/24 11:40 AM, Joost Roeleveld wrote: > >>> Those steps do not just work. > >>> The news item actually specifically states that portage will "just do > >>> the update" if you have not set any python_targets stuff. > >>> I have those not set, but it fails on ALL my systems. > >> > >> > >> Certainly it did NOT say that. > >> > >> The news item said portage would try, and indeed it did try. > > > > And failed miserably. > > I run STABLE to avoid random build failures due to library compatibility > > issues like this. > > The devs pushed a change that caused 280+ packages (Check BGO if you > > don't believe me) to simply not be compatible anymore. > > > Note that it's not a build failure -- it is an upgrade calculation > failure. It fails before upgrading any packages since it knows it can't > resolve the dependencies. > Joost also claimed that this update broke their system, which is false. If the build never started, the system is unchanged. One of three things can happen here: 1. Joost changes the defaults as explained in the news item. The Python packages don't rebuild and the state of @world is as if the upgrade never happened. 2. Joost uses autounmask to revert the change for packages that still need 3.11 support. The system is still not broken after this — in fact, a few packages just got promoted to a more modern version of Python. 3. Joost waits it out by running `emerge --sync` every day but not doing any upgrades until all conflicts are resolved. This is not the best solution, but their system is still not broken throughout this. Joost should probably subscribe to gentoo-announce in order to be notified about security vulnerabilities, though. >> You can criticize the resulting failure without claiming the news item > >> was untruthful. Thanks. > > > > The newsitem is 100% USELESS. The devs should NOT push this through with > > this MANY packages failing to even WORK > > > The purpose of the news item was to warn users that the event would > occur. Did it not achieve its purpose? Were you not-warned? > > Do we at least agree that the news item was truthful? > I found the news item quite helpful. I can't think of a way it could've been improved, can @Joost? >> Can you please tone down your complaints? Talking about how it has been > >> "forced" is both overly dramatic, and pointlessly dramatic. > > > > WHY? > > The devs FORCED a change through which caused ALL systems to FAIL > > regular maintenance. > > > This is provably false, since at least one system -- mine -- succeeded > just fine. I rebuilt several hundred packages (or synced a lot of them > from the binhost but same difference) for this upgrade, and all of them > worked. > I'd like to add that I had 157 rebuilds, with only one package being held back. That package is sys-devel/clang-15, which is the oldest version of Clang in the tree and is depended on by just one package on my system.
