Closed lloeki closed 1 year ago
Well... this seems to affect node 16.17.0 too. So either:
Just to be thorough, also fails when building with clang on Linux.
Is this caused due to compile time or runtime? can we compile in en mode and only do local stuff at runtime?
I have absolutely no idea. I have a baseless gut feeling there's some string handling problem or something. I'll try to:
16.10.0
again and see if I can reproduce (16.10.0
is known to have worked)mini_racer
(I recall there were a bunch of internal changes that led to 0.6.3
).That could help pinpointing whether it's a libv8 || node change or a mini_racer change.
Note to people watching these issues & PRs-- We are trying to release new versions and would love your help testing for stability in your products. Here's the main issue thread where we are tracking this: https://github.com/rubyjs/mini_racer/issues/277
Please let us know the results of your tests in our branches so we can do the actual releases on those branches with confidence.
With 0.8.0 released, I'm going ahead and close this one.
While attempting to evaluate #232 I hit an issue on Linux: either
es-MX
is claimed to be missing, or when replacing it withfr-FR
it produces a segfault.The failure (
es-MX
) + crash (fr-FR
) happened locally on these platforms (which were native GCC builds on their respectivex86_64
andaarch64
CPUs):The failure (
es-MX
) / crash (fr-FR
) happened in CI on these platforms (native GCC):Neither the failure nor the crash happen in CI on these platforms (which were cross-compiled by GCC), producing only a value mismatch when the locale value is changed (expected:
es-MX
does not return leading zeroes whereasfr-FR
does):All Ruby versions are affected. Moving from 17.3.1 to 17.9.1 did not help.
Crash log: