Sage 10.9.rc2 released

[Volker Braun]

As always, you can get the latest beta version from the "develop" git branch. Alternatively, the self-contained source tarball is at http://www.sagemath.org/download-latest.html

8592858fd4c (tag: 10.9.rc2, github/develop) Updated SageMath version to 10.9.rc2
19d16e0f621 gh-42083: Update PNGLIB variable to include LDFLAGS
41f3437915e gh-42035: more consistent NotImplementedErrors in hyperelliptic curves
b3600307cbf (tag: 10.9.rc1) Updated SageMath version to 10.9.rc1

[John H Palmieri]

With OS X on Apple Silicon, lots of homebrew packages:

   % make distclean && ./bootstrap && ./configure && make

fails to build fflas-ffpack. On the other hand,

   % make distclean && ./bootstrap && ./configure --with-darwin-accelerate && make testlong

works with just the expected test failures (env.py, which should be fixed by #41626, and limit_valuation.py, which has been mentioned before).

Should --with-darwin-accelerate be the default for this kind of machine? I think that's the only way to build the develop branch on OS X right now.

[Dima Pasechnik]

On Mon, May 4, 2026 at 1:28 PM John H Palmieri <jhpalm...@gmail.com> wrote:
>
> With OS X on Apple Silicon, lots of homebrew packages:
>
> % make distclean && ./bootstrap && ./configure && make
>
> fails to build fflas-ffpack. On the other hand,
>
> % make distclean && ./bootstrap && ./configure --with-darwin-accelerate && make testlong

Is this with fflas-ffpack built from source, or with Homebrew's
Macaulay2 tap for fflas-ffpack and its relatives (givaro and linbox)?

The latter use Accelerate, in this sense it's fine.
I added these taps to our lists of packages to be installed in
https://github.com/sagemath/sage/pull/42106

As far as I know, Accelerate is getting popular on macOS since it was
brought into a state compatible with recent BLAS/LAPACK releases, so
it seems OK to me to make it the default.
Packages such as scipy which used to require OpenBLAS are now working
fine with Accelerate, and ship binary wheels using Accelerate, too.

Dima

>
> works with just the expected test failures (env.py, which should be fixed by #41626, and limit_valuation.py, which has been mentioned before).
>
> Should --with-darwin-accelerate be the default for this kind of machine? I think that's the only way to build the develop branch on OS X right now.
>
>
>
>
>
> On Saturday, May 2, 2026 at 10:53:06 AM UTC-7 Volker Braun wrote:
>>
>> As always, you can get the latest beta version from the "develop" git branch. Alternatively, the self-contained source tarball is at http://www.sagemath.org/download-latest.html
>>
>>
>> 8592858fd4c (tag: 10.9.rc2, github/develop) Updated SageMath version to 10.9.rc2
>> 19d16e0f621 gh-42083: Update PNGLIB variable to include LDFLAGS
>> 41f3437915e gh-42035: more consistent NotImplementedErrors in hyperelliptic curves
>> b3600307cbf (tag: 10.9.rc1) Updated SageMath version to 10.9.rc1
>

> --
> You received this message because you are subscribed to the Google Groups "sage-release" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to sage-release...@googlegroups.com.
> To view this discussion visit https://groups.google.com/d/msgid/sage-release/83583549-1d55-4217-a5cf-86343167afe8n%40googlegroups.com.

[John H Palmieri]

This is with Sage's fflas-ffpack built from source.

[John H Palmieri]

Is there a PR to advertise "Homebrew's Macaulay2 tap for fflas-ffpack and its relatives (givaro and linbox)"? By "advertise," I mean that ./configure could print messages about these packages.

[Dima Pasechnik]

On Mon, May 4, 2026 at 3:09 PM John H Palmieri <jhpalm...@gmail.com> wrote:
>
> Is there a PR to advertise "Homebrew's Macaulay2 tap for fflas-ffpack and its relatives (givaro and linbox)"? By "advertise," I mean that ./configure could print messages about these packages.

Yes, it's https://github.com/sagemath/sage/pull/42106

(it also advertises more taps developed by me and Tobias)

> To view this discussion visit https://groups.google.com/d/msgid/sage-release/c0a6e5d2-47d9-4e6f-beda-1b368c36fb8bn%40googlegroups.com.

[John H Palmieri]

Yes, sorry, I see it there.

[John H Palmieri]

Okay, I've installed those new homebrew packages, and now I'm a little confused. In config.log I see (for example):

## ------------------------------------------------------- ##
## Checking whether SageMath should install SPKG givaro... ##
## ------------------------------------------------------- ##
real_configure:28618: checking for givaro >= 4.2.0
real_configure:28625: $PKG_CONFIG --exists --print-errors "givaro >= 4.2.0"
real_configure:28628: $? = 0
real_configure:28642: $PKG_CONFIG --exists --print-errors "givaro >= 4.2.0"
real_configure:28645: $? = 0
real_configure:28683: result: yes
real_configure:28692: will use system package and not install SPKG givaro

and then later on

real_configure:126735: result: givaro:                         no suitable system package; standard, SPKG version 4.2.1.p1 will be installed

The same happened with fflas-fpack.

Merging #42098 seems to have fixed this, but why is ./configure getting confused like this? 

[Dima Pasechnik]

On Mon, May 4, 2026 at 3:28 PM John H Palmieri <jhpalm...@gmail.com> wrote:
>
> Okay, I've installed those new homebrew packages, and now I'm a little confused. In config.log I see (for example):
>
> ## ------------------------------------------------------- ##
> ## Checking whether SageMath should install SPKG givaro... ##
> ## ------------------------------------------------------- ##
> real_configure:28618: checking for givaro >= 4.2.0
> real_configure:28625: $PKG_CONFIG --exists --print-errors "givaro >= 4.2.0"
> real_configure:28628: $? = 0
> real_configure:28642: $PKG_CONFIG --exists --print-errors "givaro >= 4.2.0"
> real_configure:28645: $? = 0
> real_configure:28683: result: yes
> real_configure:28692: will use system package and not install SPKG givaro
>
> and then later on
>
> real_configure:126735: result: givaro: no suitable system package; standard, SPKG version 4.2.1.p1 will be installed
>
> The same happened with fflas-fpack.
>
> Merging #42098 seems to have fixed this, but why is ./configure getting confused like this?

That's because linbox spkg-configure.m4 has logic which makes
givaro+fflas_ffpack rejected as soon as
linbox is rejected:

sage_spkg_install_fflas_ffpack=$sage_spkg_install_linbox
sage_spkg_install_givaro=$sage_spkg_install_linbox

(to remedy some incompatibility things IIRC)

To let linbox be accepted you need Homebrew taps for
all its deps: fflas_ffpack flint fplll givaro gmp iml m4ri m4rie mpfr ntl

If you really install all the taps from #42106, or at least m4ri,
m4rie, and iml (and uninstall the corresponding Sage-installed
packages, obviously) then it would work

Dima

> To view this discussion visit https://groups.google.com/d/msgid/sage-release/95676772-8f81-4726-b7e5-4112d4dbd91fn%40googlegroups.com.

[John H Palmieri]

Okay, thanks, although config.log could be clearer about what's going on; for example, the linbox section could tell us that it's also rejecting givaro and fflas-fpack, so that a search through that file would explain what's going on.

In the meantime, with #42098 and the new homebrew packages, the build (without using '--with-darwin-accelerate') fails on sagelib with the same error reported at #41626: error: unknown argument: '-MD'.

If I use '--with-darwin-accelerate', the build also fails on sagelib. I see this error: "/opt/homebrew/Cellar/givaro/4.2.1_1/include/recint/rrint.h:125:32: error: 'make_signed' cannot be specialized: Users are not allowed to specialize this standard library entity [-Winvalid-specialization]"

[Dima Pasechnik]

On Mon, May 4, 2026 at 4:07 PM John H Palmieri <jhpalm...@gmail.com> wrote:
>
> Okay, thanks, although config.log could be clearer about what's going on; for example, the linbox section could tell us that it's also rejecting givaro and fflas-fpack, so that a search through that file would explain what's going on.
>
> In the meantime, with #42098 and the new homebrew packages, the build (without using '--with-darwin-accelerate') fails on sagelib with the same error reported at #41626: error: unknown argument: '-MD'.

That's, I suppose, the same Meson bug as discussed on #41626

>
> If I use '--with-darwin-accelerate', the build also fails on sagelib. I see this error: "/opt/homebrew/Cellar/givaro/4.2.1_1/include/recint/rrint.h:125:32: error: 'make_signed' cannot be specialized: Users are not allowed to specialize this standard library entity [-Winvalid-specialization]"

That's because Homebrew's givaro tap does not yet have
https://github.com/linbox-team/givaro/pull/240
I wonder why I don't see this error on my machine, though. I have

Apple clang version 21.0.0 (clang-2100.0.123.102)
Target: arm64-apple-darwin25.4.0

I'll do a PR to fix Macaulay2 givaro tap, hopefully it'll quickly gets in.

> To view this discussion visit https://groups.google.com/d/msgid/sage-release/8e40b576-c756-4b09-9ba3-fedf43fbfff6n%40googlegroups.com.