Author glaubitz
Recipients Arfrever, David.Edelsohn, christian.heimes, cstratak, glaubitz, lukasz.langa, vstinner
Date 2021-02-16.08:32:56
SpamBayes Score -1.0
Marked as misclassified Yes
Message-id <1613464377.33.0.4456925242.issue43179@roundup.psfhosted.org>
In-reply-to
Content
> This thread is an excellent example why ignoring platforms comes at a cost. It will only get worse when are going to introduce platform and architecture specific code for optimizations and features.

Which is purely hypothetical at the moment. You are arguing with something that might happen in the future but currently doesn't exist to justify the removal of 5 lines of preprocessor and autoconf "code".

You can still drop these lines in the future _if_ they happen to cause any headache. But that is currently not the case, so there isn't really a burden.

>> You can view test results any time by going to https://buildd.debian.org/ and searching for "pythonX.Y". So there is actually a CI for release builds.

> The site does not list a s390 builder. There is only a s390x builder.

s390 is being built for SLE-12, for example, on the internal SUSE build system and SLE-12 is still supported. So if a customer wants to use Python 3.10 in a SLE-12 s390 environment, why keep them from doing so?

In my experience some upstream projects make the mistake that they think they always know how users are using their software. But since there is no dedicated feedback channel, you have no means in knowing whether someone is building Python for a given architecture for their custom project. After all, there are source-only distributions like Gentoo, so you don't have to rely on any existing binary builds.
History
Date User Action Args
2021-02-16 08:32:57glaubitzsetrecipients: + glaubitz, vstinner, christian.heimes, Arfrever, lukasz.langa, David.Edelsohn, cstratak
2021-02-16 08:32:57glaubitzsetmessageid: <1613464377.33.0.4456925242.issue43179@roundup.psfhosted.org>
2021-02-16 08:32:57glaubitzlinkissue43179 messages
2021-02-16 08:32:56glaubitzcreate