Have you tested if sage still builds after this update?

@timokau No. I thought the CI does that?

No, ofBorg only builds the package itself by default. Building all reverse-dependencies of every PR would take too much resources. Its possible to tell ofBorg to build additional packages, but in this case it would most likely time out.

(Edit: I submitted a PR with this change to your fork)

According to https://trac.sagemath.org/ticket/30985, you probably need to apply a small patch available at https://git.sagemath.org/sage.git/patch?id2=b16b6b5b0c815482560b00d12f7e528d7aab1dcd&id=de5d173ede03cd4932661dbafaf1e02f927f64d1. Typically you would use the fetchSageDiff function, but since the patch does not apply cleanly to the 9.2 branch, here's an untested rebased version that you can add to the patch list at https://github.com/NixOS/nixpkgs/blob/master/pkgs/applications/science/math/sage/sage-src.nix#L95

diff --git a/src/sage/interfaces/sympy.py b/src/sage/interfaces/sympy.py
index cc35a42a9f..6e577d5d8d 100644
--- a/src/sage/interfaces/sympy.py
+++ b/src/sage/interfaces/sympy.py
@@ -397,7 +397,7 @@ def _sympysage_rf(self):
         sage: from sympy import Symbol, rf
         sage: _ = var('x, y')
         sage: rfxy = rf(Symbol('x'), Symbol('y'))
-        sage: assert rising_factorial(x,y)._sympy_() == rfxy.rewrite('gamma')
+        sage: assert rising_factorial(x,y)._sympy_() == rfxy.rewrite('gamma', piecewise=False)
         sage: assert rising_factorial(x,y) == rfxy._sage_()
     """
     from sage.arith.all import rising_factorial
diff --git a/src/sage/symbolic/expression.pyx b/src/sage/symbolic/expression.pyx
index 7c18ec1efa..c2619ac42d 100644
--- a/src/sage/symbolic/expression.pyx
+++ b/src/sage/symbolic/expression.pyx
@@ -955,7 +955,7 @@ cdef class Expression(CommutativeRingElement):
             sage: unicode_art(13 - I)
             13 - ⅈ
             sage: unicode_art(1.3 - I)
-            1.3 - 1.0⋅ⅈ
+            1.3 - ⅈ
             sage: unicode_art(cos(I))
             cosh(1)

Let me know if you encounter any issues.

Note that the singular.pyx Sage segfault is intermittent, doesn't happen too often, happens upstream too and is unfixed there (https://trac.sagemath.org/ticket/30945). If that's the only failure (which was the case on the x86_64 builder), I think we can consider the test run a success.

Edit: Looks like the aarch64 build was killed. It had to build a jdk and R for some reason, so it probably timed out before it could run the tests. If those outputs are cached it should run fine next time, let's see.

@ofborg build sage

Yep, this new run was all green, both on aarch64 and on x86_64. Sage seems OK with the update.

Note that the singular.pyx Sage segfault is intermittent

In that case I think it would be best to disable that test. Otherwise users will have to run the whole testsuite locally whenever the test fails on hydra.

That is unrelated to this PR though. Thanks for the reviews @collares and @omasanori!

Took the liberty to squash & merge to avoid an intermediate commit where sage is broken.