Gonzalo Tornaría

> @tornaria if you can test the current commit [d209a2d](https://github.com/JohnCremona/eclib/commit/d209a2df9d0dce6c50e91d90d1ebf448d12a75cf) please would you do so and report here? I think it is better, pending more work to use a library...

If you apply this: ``` --- a/libsrc/interface.cc +++ b/libsrc/interface.cc @@ -125,8 +125,8 @@ void Compute_Euler(RR& y) add(u, u, a); add(v, v, b); } - div(y, u, v); RR::SetPrecision(l); + div(y,...

I don't understand what's wrong with the code. It's process substitution, is that not allowed in your version of bash? Note that this feature is also used in make completion....

Sorry, my bad. I see now it's the same failure you found before, I thought it was a failure in the test framework. In fact if I revert my patch...

Rebased on top of 2.7. The first commit 944a507 is only about expanding the test suite to be more comprehensive, including two tests that currently fail (one of them corresponds...

Rebased on top of #50295, updated to sagemath 10.4.beta6

@dkwo sagemath 10.4 is getting close, this works fine for me on x86_64 and passes all tests on i686 and x86_64-musl. If you want to check on aarch64 this is...

Ideally, this PR should be merged on top of #50565, so that gsl is updated beforehand (both giac + sagemath need to be rebuilt when gsl is updated). Edit: I've...

> Thanks. I won't have access to my aarch64 machines for a month or so. I think this should be merged now, and I'll run tests once I get back....

I tested locally on x86_64 with numpy 2.0 as follows: - On top of this PR I merged #51077 (at a656cfb6837 to be more precise). - From the numpy branch,...