On Monday, October 11, 2021 at 11:20:41 AM UTC-7 Nils Bruin wrote: > Another worry that I have is that modularizing sage will lead to > destabilization. The original reason for putting all components of sagemath > into one big environment was to reduce the number of variables in getting > things to work. I expect that if modularization takes off and sage ends up > split in more components, then 10 years down the road there will be a group > of developers that throw their hands up in desperation because it's just so > hard to get all different components together with versions that actually > work well together, and trying to keep different components working across > version ranges is just so tricky (plus it's too hard to get exactly the > right versions from all places, because the newest version of some > components will always just have made some incompatible change that the > others still have to catch up to). So PLEASE make it easy to have a > reference of known-to-work components together, like sage-the-distribution > does. I think we'll need it in the future again. >
I fully agree with all of this, and it matches my plan and recommendations. I find it helpful to think about *sage-the-distribution (which is more or less the same as the contents of the Sage monorepo) as the reference distribution for Sage developers*, in which, as Nils says, all the components fit together, and developers are able to make changes without having to buy into the specifics of packaging systems such as nix, conda, ... To me it is secondary whether we advertise Sage-the-distribution to end users -- in fact, there may be better distribution options (such as conda) for end users. -- 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/2b823307-e9a1-4522-9fa7-4ccca09a8e00n%40googlegroups.com.
