singular: update to 4.3.1p2 (fix gcc 12 build)

[dkwo]

I tested the changes in this PR: built and checked on x86_64 using the gcc 12 PR.
(Even the non-musl version had the same failure with gcc 12.)
Had to use patchversion p2 otherwise it would still fail (with different errors).
@tornaria @leahneukirchen @Duncaen

[tornaria]

Just as a heads up: I'm working on updating sage to 9.7. Maybe it's worth doing all in one swipe?

[dkwo]

Sure, that's fine with me. Just incorporate this one there? Also perhaps have a look at the python3.11 update, I think it's also updating sage.

[tornaria]

Sure, that's fine with me. Just incorporate this one there?

Sure, I will.

Just FYI, a few doctest failures in sagemath 9.5 are expected, mostly safe to ignore. Right now I have all doctests passing with sagemath 9.7 + a few patches (less than in 9.5). However, that's using matplotlib 3.5.3. I haven't yet started working on matplotlib 3.6 (which causes most of the failures you have here, #39857 is a first step that I'm yet to test, but I'm not sure that will be enough).

I will include this in my sagemath PR, together with other updates I've been doing (some of which work ok independently of sagemath, others like singular need rebuild of sagemath).

Also perhaps have a look at the python3.11 update, I think it's also updating sage.

Sure, but locking sagemath update to python update seems more difficult. OTOH locking singular and sagemath seems more reasonable (unless you have a pressing reason to update singular, if so don't wait for me).

Anyway, maybe worth testing the sagemath update with python 3.11 just to make sure everything works.

BTW, the stupid limit recent python imposed on "int(s)" is also causing a few failures but they will go away with sagemath 9.7 (they just disable the limit).

[dkwo]

I'm not in a hurry, the main reason I wanted to update is for the gcc12 update, which fails at building previous singular versions. Please link your sage PR here once available, and I'll close this.

[tornaria]

BTW, you could do something like this:

diff --git a/srcpkgs/singular/template b/srcpkgs/singular/template
index 441a5ba561..824345b27f 100644
--- a/srcpkgs/singular/template
+++ b/srcpkgs/singular/template
@@ -1,9 +1,8 @@
 # Template file for 'singular'
 pkgname=singular
-_patchver=p2
-_majver=4.3.1
-version=${_majver}${_patchver}
+version=4.3.1p2
 revision=1
+_majver=${version%p*}
 wrksrc=singular-${_majver}
 build_style=gnu-configure
 configure_args="--with-readline=ncurses

to avoid the xlint complain. I believe it is mandatory to have the full expanded version string in 'version=...' so that it can be easily parsed with grep and sed from templates (just grep "^version=", etc.). The ${version%p*} thing should work ok for versions with a patchlevel as for versions without it.

[tornaria]

Also:

srcpkgs/singular/template:9: variable set continuation should start with a single space
srcpkgs/singular/template:10: variable set continuation should start with a single space
srcpkgs/singular/template:11: variable set continuation should start with a single space
srcpkgs/singular/template:12: variable set continuation should start with a single space
srcpkgs/singular/template:13: variable set continuation should start with a single space
srcpkgs/singular/template:14: variable set continuation should start with a single space
srcpkgs/singular/template:15: variable set continuation should start with a single space
srcpkgs/singular/template:16: variable set continuation should start with a single space

but that's with a patch to xlint that I've just realized I didn't submit upstream 🤦

Edit: it's now leahneukirchen/xtools#266.

[dkwo]

Done.

[leahneukirchen]

Do you want this merged now or as part of later Sage update?

[tornaria]

Do you want this merged now or as part of later Sage update?

It's a matter of taste: merging this now will force rebuild of sagemath-9.5. Then the update to sagemath-9.7 will force rebuild of this (due to updating flintlib, which is a dependency of singular). I think the main interest of this update is to support gcc12.

I expect to be done with sagemath-9.7 by the weekend, so I'd rather wait and do all together.

[leahneukirchen]

Ok. As long as we don't need to wait for Python 3.11.

[dkwo]

Sure, let's do it in sagemath pr then.