Sirocco update to a pkg-config enabled version

[dimpase]

also fix a typo in spkg-configure.m4 of roman_numerals_py, and fix spkg-configure.m4 of flint to allow versions 3.3.* (will be done elsewhere - in #40568)

📝 Checklist

• The title is concise and informative.
• The description explains in detail what this PR is about.
• I have linked a relevant issue or discussion.
• I have created tests covering the changes.
• I have updated the documentation and checked the documentation preview.

⌛ Dependencies

[dimpase]

@tobiasdiez - not sure, is the following the good fix for the meson build:

--- a/src/sage/libs/meson.build
+++ b/src/sage/libs/meson.build
@@ -1,4 +1,3 @@
-sirocco = cc.find_library('sirocco', required: false, disabler: true)
 # cannot be found via pkg-config
 ecl = cc.find_library('ecl', required: false, disabler: true)
 if not ecl.found() and not is_windows
@@ -65,6 +64,12 @@ else
   conf_data.set('SAGE_MAXIMA', maxima_bin.full_path())
 endif
 
+sirocco = dependency(
+  'libsirocco',
+  version: '>=2.1.0',
+  required: false,
+  disabler: true,
+)
 braiding = dependency(
   'libbraiding',
   version: '>=1.3.1',

[dimpase]

sirocco tarball is an unreleased version I made from my Sirocco PR miguelmarco/SIROCCO2#7

[github-actions]

Documentation preview for this PR (built with commit a6827d0; changes) is ready! 🎉
This preview will update shortly after each push to this PR.

[kiwifb]

Is the .pc file needed for the meson build?

[dimpase]

Is the .pc file needed for the meson build?

yes, and I checked that with the last commit meson build of sirocco interface works

[antonio-rojas]

@tobiasdiez - not sure, is the following the good fix for the meson build:

--- a/src/sage/libs/meson.build
+++ b/src/sage/libs/meson.build
@@ -1,4 +1,3 @@
-sirocco = cc.find_library('sirocco', required: false, disabler: true)
 # cannot be found via pkg-config
 ecl = cc.find_library('ecl', required: false, disabler: true)
 if not ecl.found() and not is_windows
@@ -65,6 +64,12 @@ else
   conf_data.set('SAGE_MAXIMA', maxima_bin.full_path())
 endif
 
+sirocco = dependency(
+  'libsirocco',
+  version: '>=2.1.0',
+  required: false,
+  disabler: true,
+)
 braiding = dependency(
   'libbraiding',
   version: '>=1.3.1',

the meson build works just fine with the current release of sirocco (without pkgconfig). Please don't make it depend on an unofficial fork.

[dimpase]

it's not an unofficial fork, it's just a temporary fix, until @miguelmarco merges my PR

[antonio-rojas]

it's not an unofficial fork, it's just a temporary fix, until @miguelmarco merges my PR

Still, until it is merged and released this will break distros, including conda (you can see in the conda testers that sirocco is not detected). Please revert the meson bits, there is nothing broken that needs fixing there.

[dimpase]

It didn't work for me without pkg-config for libsirocco in SAGE_LOCAL.

[tobiasdiez]

You can first try via pkg-config and if that's not working fallback to find_library similar to how its done for gap:
https://github.com/sagemath/sage/blob/858268b40010e5ed6da13488ad0f52cda4d1f70e/src/meson.build#L191C1-L201C6

[orlitzky]

You can first try via pkg-config and if that's not working fallback to find_library similar to how its done for gap: https://github.com/sagemath/sage/blob/858268b40010e5ed6da13488ad0f52cda4d1f70e/src/meson.build#L191C1-L201C6

Might want to rebase over #40485 as well since it touches this line.

[dimpase]

OK, I've reverted the commit touching meson.build. It should be not so controversial now.

[enriqueartal]

I uninstalled both sirocco and sagemath_sirocco, applied the patch, ./configure --enable-sirocco, make build and got this:

┌────────────────────────────────────────────────────────────────────┐
│ SageMath version 10.7.rc2, Release Date: 2025-08-09                │
│ Using Python 3.13.5. Type "help()" for help.                       │
└────────────────────────────────────────────────────────────────────┘
┏━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┓
┃ Warning: this is a prerelease version, and it may be unstable.     ┃
┗━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━━┛
sage: A.<x, y> = AffinePlaneCurveArrangements(QQ)
sage: C = A(y^2 - x^3)
sage: C.fundamental_group()
Exception raised by child process with pid=457157:
Traceback (most recent call last):
  File "/usr/local/sage15/src/sage/parallel/use_fork.py", line 191, in __call__
    self._subprocess(f, dir, *v0)
    ~~~~~~~~~~~~~~~~^^^^^^^^^^^^^
  File "/usr/local/sage15/src/sage/parallel/use_fork.py", line 328, in _subprocess
    value = f(*args, **kwds)
  File "/usr/local/sage15/src/sage/schemes/curves/zariski_vankampen.py", line 898, in braid_in_segment
    aux = followstrand(f, [p for p in glist if p != f],
                       x0, x1, i.center(), precision1[f])
  File "/usr/local/sage15/src/sage/schemes/curves/zariski_vankampen.py", line 567, in followstrand
    from sage.libs.sirocco import (contpath, contpath_mp, contpath_comps, contpath_mp_comps)
ModuleNotFoundError: No module named 'sage.libs.sirocco'
---------------------------------------------------------------------------
AttributeError                            Traceback (most recent call last)
Cell In[3], line 1
----> 1 C.fundamental_group()

File /usr/local/sage15/src/sage/schemes/curves/plane_curve_arrangement.py:569, in AffinePlaneCurveArrangementElement.fundamental_group(self, simplified, vertical, projective)
    567 else:
    568     bd = None
--> 569 G, dic = fundamental_group_arrangement(L, simplified=simplified,
    570                                        puiseux=True,
    571                                        projective=projective,
    572                                        vertical=vertical,
    573                                        braid_data=bd)
    574 if simplified and vertical:
    575     self._fundamental_group_simpl_vertical = G

File /usr/local/sage15/src/sage/schemes/curves/zariski_vankampen.py:1902, in fundamental_group_arrangement(flist, simplified, projective, puiseux, vertical, braid_data)
   1900     d1 = 0
   1901 else:
-> 1902     bm, dic, dv, d1 = braid_monodromy(f, flist1, vertical=vertical)
   1903 vert_lines = list(dv)
   1904 vert_lines.sort()

File /usr/local/sage15/src/sage/schemes/curves/zariski_vankampen.py:1379, in braid_monodromy(f, arrangement, vertical)
   1377         b = braidcomputed[1]
   1378         segsbraids[(beginseg, endseg)] = b
-> 1379         segsbraids[(endseg, beginseg)] = b.inverse()
   1380     end_braid_computation = True
   1381 except ChildProcessError:  # hack to deal with random fails first time

AttributeError: 'str' object has no attribute 'inverse'

The same error without the patch. After ./configure --enable-sagemath_sirocco, make build:

sage: A.<x, y> = AffinePlaneCurveArrangements(QQ)
sage: C = A(y^2 - x^3)
sage: C.fundamental_group()
Finitely presented group < x0, x1 | x1*x0*x1*x0^-1*x1^-1*x0^-1 >

as it should be. I think the key is ModuleNotFoundError: No module named 'sage.libs.sirocco'.

[dimpase]

if I do the latter, I get a non-working Sage, for some reason.

what you see is that enable-sirocco only builds the sirocco library, and the current build needs enable-sagemath_sirocco option to build the interface to sirocco.

in meson build it is automatically detected whether libsirocco is available, and the interface is built (but it is getting an option to control whether it's actually built)

[enriqueartal]

Is there a way to build the interface to sirocco with --enable-sirocco?

[dimpase]

start sage --buildsh and follow the detailed meson steps in installation/meson.rst

[enriqueartal]

I am trying, but I would like to know if it is possible when building in the classical way.

[dimpase]

it's not an unofficial fork, it's just a temporary fix, until @miguelmarco merges my PR

Still, until it is merged and released this will break distros, including conda (you can see in the conda testers that sirocco is not detected). Please revert the meson bits, there is nothing broken that needs fixing there.

@antonio-rojas - this is now based on the official new release.
I also too away unrelated changes, and don't touch meson. OK?

[antonio-rojas]

I can't test myself this month, but conda CI looks fine, so it should be ok on Arch too.

[tobiasdiez]

LGTM, but perhaps someone actually using that stuff is a better reviewer ;-)

[enriqueartal]

I am trying to build sage with meson for the first time to see if these changes work for me. I had some issues, see this message. If I get it I have some doubts:

• How is an optional package installed with meson?
• How can I force to use a sage package instead of a system one? For example, my flint system package is 3.2.2.

[dimpase]

Please, this PR is a trivial update (apart from spkg-configure.m4, which you can only test on a system which has system-wide sirocco, or sirocco installed by hand in /usr/local or so)