On Thu, 2021-09-23 at 23:17 +0100, Dima Pasechnik wrote:
> https://trac.sagemath.org/ticket/32532
> <https://trac.sagemath.org/ticket/32532#comment:21> proposes to remove
> these packages as not needed, and a huge time sink for everyone involved.
> 

I'm sure this list changes every time I make it up, but here are the
pros/cons that I see:

Pro (in favor of removal):

  1. gcc and gfortran are available everywhere; and the (recent) 
     default behavior is to exit ./configure with an error if the
     user does not have gcc installed on his system anyway.

  2. It is always an error for end-users to install these SPKGs.
     If you have a problem installing your "native" gcc/gfortran
     packages, then we (on this list) will tell you how to solve
     that problem; not to use the SPKG.

  3. If all else fails, using homebrew, conda, nix, or something
     similar is still preferable to the SPKG.

  4. As you note, they require a bunch of developer time that is
     more or less wasted, since no one should ever use them.

  5. They're standard packages, so the (large) sources are shipped
     as part of the (thus even larger) release tarballs.

And the cons (in favor of keeping them):

  1. Inertia. If someone has these as part of his workflow, us removing
     them is going to cause him some headaches, despite whatever moral
     high ground we can claim.

  2. As a truly last resort, anyone with a sage tree can try to
     build the SPKG without having to learn conda or nix or how
     to do a manual install into /usr/local.

  3. Removing them means we'll have to re-think the detection that 
     currently happens in the corresponding spkg-configure.m4 files.

  4. Having them as SPKGs means that I can create a branch that 
     upgrades the gcc/gfortran packages, and use Github actions
     to easily test sage with future releases.

I think the main value is in (4), but personally it's not enough for
me. Testing a new gcc/gfortran is easy to do in a Gentoo container,
where you can just upgrade to the new gcc/gfortran, select it, and then
try to build sage. That won't let you test brand-new-gcc-on-an-ancient-
ubuntu to find unrelated ubuntu bugs, but those should not be the
responsibility of the sage developers. You'll be using "apt" to install
gcc/fortran on ubuntu anyway.

-- 
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/9cf3fe09f4c859589ed1b38d51e2d23e1b5d71e0.camel%40orlitzky.com.

Reply via email to