On Wed, 2023-04-26 at 13:06 -0700, Matthias Koeppe wrote:
>
> 2. I'm not in favor of chipping away 1 package at a time in the name
> of unsubstantiated, vague notions that a package is "ballast slowing
> down Sage's progress".
>
There's a ticket open to update PARI within Sage. First, upstream had
to release the new version. So far, at least three sage developers have
worked on it. The update includes a backported patch, a test for the
patch, and configure checks for the patch. So, once the update makes
its way into sage, every distribution maintainer is going to have to
backport the same patch into their version of PARI. Except the sage
patch doesn't include the fix to the test suite. So the source based
distributions are also going to have to separately backport the fix for
the test suites. There are 12 distros listed in build/pari/distros, so
at least 12 people this will affect.
How many people will it take how many hours to complete this trivial
upgrade? These are not vague notions.
Dropping one package at a time has two major benefits:
1. Not all sage packages are available in all distros. Eliminating
the easy ones like gcc and python makes it possible to focus on
the ones that pose real problems.
2. The sage library and test suites are coupled much too tightly to
dependencies like PARI and maxima. Piggybacking off the earlier
example, distroless Sage cannot continue to test for patches that
exist only in upstream git head. We should also endeavour to make
our tests independent of the particular string representation that
e.g. maxima uses. There are lots of tests that work only with one
specific version of maxima because the answers look different,
rather than because they are fundamentally unequal. Fixing all of
these issues will take a lot of time and work. It will only be
feasible to fix them if we do it a little bit at a time. We should
fix them anyway because updating the tests every time a new maxima
is released is a waste of time. But we must fix them if we expect
the distro maxima to work.
--
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/29a6ceea545aaea4406dc7ce294e1c87943b5d1b.camel%40orlitzky.com.
