Nathan explained it correctly. On Tuesday, August 6, 2024 at 1:45:21 PM UTC-7 Nathan Dunfield wrote:
> Dima, > > As I understand it, while both proposals result in binary wheels being > pulled off PyPI for certain standard packages, they differ in that: > > a) The current proposal requires the version of the package to be pinned > (with checksums). > > b) The current proposal does not allow a package to pull in any > dependencies that are not explicitly included as other Sage packages. > > As you know, these are both part of the behavior of the current "wheel > packages", which are currently required to be pure Python, with no compiled > extensions. > > In contrast, our "pip packages", used for so many optional packages, do > not require either (a) or (b), and I think most do not. Your proposal, as > I understood it, was to allow standard packages to that do not do either > (a) or (b). My comment you refer to is my reasons for why I think (a) and > (b) are important. Later in that same thread, said I was in favor of Sage > using upstream binary wheels from PyPI if we did (a) and (b): > > https://groups.google.com/g/sage-devel/c/5kmxaw105lg/m/fu-g4zhaEwAJ > > Best wishes, > > Nathan > > > On Tuesday, August 6, 2024 at 3:17:15 PM UTC-5 Dima Pasechnik wrote: > >> Nathan, >> this is essentially the proposal I put forward in Feb 2024: >> https://groups.google.com/g/sage-devel/c/5kmxaw105lg/m/QGShvy6-AAAJ >> >> You didn't like it then. >> https://groups.google.com/g/sage-devel/c/5kmxaw105lg/m/N-eEEQTeAgAJ >> >> Dima >> >> PS. Nobody seems to remember that conversation? Is it forbidden to >> mention my proposals on sage-develop nowadays? >> >> On Tue, Aug 6, 2024 at 1:49 PM Nathan Dunfield <nat...@dunfield.info> >> wrote: >> > >> > This proposal makes sense to me. Something like this is necessary for >> us to upgrade Jupyter since it depends on rdps-py which needs Rust to >> build, and this approach preserves the other features of current standard >> packages: the version is pinned (including checksums on the binary wheels) >> and all dependencies explicitly included as other standard packages. >> > >> > Nathan >> > >> > On Sunday, August 4, 2024 at 1:03:38 PM UTC-5 Matthias Koeppe wrote: >> >> >> >> Currently all "standard" packages of the Sage distribution, by policy ( >> https://doc.sagemath.org/html/en/developer/packaging.html; recommended >> reading), >> >> - either can be installed from source ("normal" packages); >> >> - or they are Python packages that can be installed from >> platform-independent wheels. >> >> >> >> (We added the second option in order to treat Javascript assets as >> prebuilt data, thus avoiding the complexity of Javascript build >> infrastructure in the Sage distribution.) >> >> >> >> In https://github.com/sagemath/sage/pull/38219 (needs review), I >> propose the following mild policy change: >> >> - as a third option, a "standard" package is allowed to be a Python >> package that can be installed from platform-dependent (binary) wheels. In >> this case, a "configure" option must be provided that disables the package. >> >> >> >> (Again this new option is motivated by avoiding the complexity of new >> build infrastructure related to another language; in this case, Rust. See >> the PR for two example Rust-based example packages; previous discussion: >> https://groups.google.com/g/sage-devel/c/JiSryImYGe0/m/4B8z6F4AEwAJ) >> >> >> > -- >> > 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+...@googlegroups.com. >> > To view this discussion on the web visit >> https://groups.google.com/d/msgid/sage-devel/b1ccf359-c759-46e7-a08a-f1176d91f755n%40googlegroups.com. >> >> >> > -- 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/a7fbf965-03fd-4af1-b909-5ef51db6e60bn%40googlegroups.com.
