On Thu, 2023-06-08 at 14:09 -0700, Matthias Koeppe wrote: > > *D. *As a consequence of B and C, it was *impossible to build or run parts > of the Sage library.* And it is *impossible to install the whole Sage > library using Python infrastructure* (pip). (Yes, I know that conda exists.) >
Of itself, modularization is a noble goal. But in the implementation, please be wary of focusing on users we don't have at the expense of the users we do have. The amount of additional complexity and maintenance this entails is daunting, and your tentative plans for pip packaging sometimes conflict with common-sense improvements for existing users and developers, as I've been trying to argue (for the second time) on issue #32242. Pip isn't a real build system or package manager. Most egregiously, it can't do anything with the non-python software on which sage subsists. To make sage-via-pip work, we'll have to maintain a new pseudo- distribution on pypi that either ships people pre-built wheels or wraps autotools/cmake/etc in python. As was made clear in recent threads, many developers don't want to be maintaining the *first* sage distribution, much less a second one. I think we're all happy to modularize so long as it doesn't create work elsewhere. (At the very least, modularization improves build times.) But given that you may be the only person who has pip-installablity as a goal, I think you have to be more sensitive to the complaints when that goal conflicts with others. Complexity is our biggest problem right now, and uncompromising modularization is only adding to it. -- You received this message because you are subscribed to the Google Groups "sage-devel" group. To unsubscribe from this group and stop receiving emails from it, send an email to sage-devel+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/sage-devel/f6cc06ddf559985a33303529d45b1eaaf9b1cf98.camel%40orlitzky.com.
