Author ncoghlan
Recipients Arfrever, barry, brett.cannon, doko, eric.snow, lemburg, ncoghlan, ned.deily, pitrou, python-dev, steve.dower, tim.golden, zach.ware
Date 2015-04-16.16:12:03
SpamBayes Score -1.0
Marked as misclassified Yes
Message-id <1429200723.84.0.535623861234.issue22980@psf.upfronthosting.co.za>
In-reply-to
Content
I can tell you exactly where these files need to live side by side: index servers. We currently paper over it on PyPI for Windows and Mac OS X by leveraging the implicitly defined ABI compatibility of the CPython binary releases published on python.org, but that starts to break down as soon as people are using interpreter binaries built locally or by redistributors rather than by the CPython core development team. There's more to the computing world than Intel CPU architectures, and more kernel APIs than Linux, Darwin and Windows.

Fortunately, this platform ABI identification issue is not a new problem, but rather one where there's a rich history of existing expertise that we can draw from, rather than inventing our own custom solution that doesn't actually solve the problem properly. I agree it needs to be documented appropriately, perhaps including a PEP to explain *why* the changes are being made (rather than just documenting the end result), which is why I opened issue 23966.
History
Date User Action Args
2015-04-16 16:12:03ncoghlansetrecipients: + ncoghlan, lemburg, barry, brett.cannon, doko, pitrou, tim.golden, ned.deily, Arfrever, python-dev, eric.snow, zach.ware, steve.dower
2015-04-16 16:12:03ncoghlansetmessageid: <1429200723.84.0.535623861234.issue22980@psf.upfronthosting.co.za>
2015-04-16 16:12:03ncoghlanlinkissue22980 messages
2015-04-16 16:12:03ncoghlancreate