Hello, On Tue, Jun 4, 2024 at 4:37 PM Joost Roeleveld <jo...@antarean.org> wrote:
> > > LAST ATTEMPT AS YOU CLEARLY ARE INCAPABLE OF UNDERSTANDING NORMAL > ENGLISH. I WILL USE CAPS FOR YOU. > > WHERE DID I STATE I WANT TO DEFER THE UPDATE? > I DO NOT CARE ABOUT THE VERSION! > > THE ISSUE IS CAUSED BY A BROKEN SET OF PACKAGE VERSION IN PORTAGE! > > reading through your emails, I understand that we share the view that switching the default to python 3.12 could have waited until more packages in stable had support for it. With this premise I'd like to make you reason with me of where this line of thought fails. I'll put my science hat on, and hope you won't be bothered by it. First fundamental rule of statistics is that we can't make population-wide statements from single local observations. Consequently, when you and I see our system fail the upgrade, we have no basis to assert that the decision of the development team to switch default was wrong. As your systems and mine could represent a very small part of the Universe of Gentoo systems that the developers need to make decisions for. On a similar note, any argument on the number of packages that are not yet python 3.12 ready is, to some extent, very weak. In fact, these summary statistics are not telling the entire story. A more proper analysis should take into account the likely-hood of these packages actually being installed on a meaningful fraction of Gentoo systems. All this information should be easily obtainable from the download statistics of the gentoo mirrors. You seem to expect that they should have changed the default only when *everything* in stable was ready for python 3.12. Let me tell you that this expectation is utopian at best. They would need to be working with infinite resources (time and money) which is not the case for anyone. Your systems, mine, and those of the people writing in here might be the unfortunate 5% of the overall total (I am using 5% as science uses that as reasonable error, the development team might have different tolerance for error). What I believe is in the realm of reasonable is to ask to be notified when important (as in popular) packages that are currently missing support get updated with a stable version supporting python 3.12 so that we can take action on our side. We can also ask, if resource are available, to provide binary packages with python 3.11 support as it makes upgrades faster for those users that couldn't bump to python 3.12. However, while we can ask, the cost-benefit ratio might not justify the additional resources needed to maintain them and, quite frankly, you and I have no voice in that decision. Best regards, Marco.
