SciPy builds are failing with gast 0.5.4

[tylerjereddy]

SciPy meson source builds work with gast 0.5.3 but not 0.5.4 per scipy/scipy#18390 (comment).

[Tatsh]

serge-sans-paille/pythran#2101 (comment)

[Tatsh]

Caused by this 0.5.3...0.5.4#diff-44b67094998a7df1e5ce843b5a2faa44950f5896ac9e07ab99d09e2b24c90143R13

[serge-sans-paille]

And fixed by serge-sans-paille/pythran#2100

[h-vetinari]

And fixed by serge-sans-paille/pythran#2100

Since that causes yet another breakage, could we take a step back and find a less stressful way to roll out this upgrade?

Alternatives I can think of:

• Backport serge-sans-paille/pythran@f0c2777 to pythran 0.12 and release 0.12.2
• Yank gast 0.5.4

Context: The CI of scipy (and other projects) has been broken for ~2 days now. This is not a good place to be and while I appreciate that regressions can happen, I'd hope that this situation could get unbroken first before we wait for the next fixes on top.