Update list of supported compilers #192
Conversation
|
Hm, am I looking at the right build here? Both master builds are green for LDC 1.9.0/1.14.0, AFAICS. There have been two failures for DMD, though. I have restarted them to see whether they are flaky. |
Geod24
force-pushed
the
update-compiler
branch
from
36c717f
to
802e718
Compare
|
Ah, now I see, are you are looking at #188? I've restarted the failing builds, but it seems like the changes actually do somehow trigger them. I haven't seen those on master and the restarted builds fail in the same way. |
|
Yes I was under the impression that it triggered on master as well, but you're right it doesn't, and the change do trigger them. There is no consistent failure, only a somewhat flakey dirwatcher test. Now looking again at #188 I can't really figure out why this set of change would trigger this bug... But that's a separate topic. Should I just reinstante 1.4.0 and 1.14.0 LDC ? |
It is deprecated and hence not required anymore.
They have been consistently failing on the CI for months, and there doesn't seem to be any interest in fixing it, however they do block new PRs.
Update the list to include the latest compilers (DMD 2.091.0 and LDC 1.20.1), as well as the previous 3 releases, plus the oldest supported release. Also switch from beta to the latest CI version, which is more informative, but still allow it to fail so as not to block development.
|
Updated with LDC 1.20.1 and DMD 2.091.0 |
|
Travis is green, but I've seen spurious failures |
|
I have restarted them, they happen on the macOS VMs due to their reliance on timing and the huge spikes that occur there. One day I'll fix them with proper synchronization primitives. |
|
Since I am developing on Mac OSX, I'd love to see that day :D |
Since LDC v1.9.0 and v1.14.0 are consistently failing, remove them from the CI.
It's not my preferred solution to remove a failing test, but no one seems interested in fixing this (master has been red for over 2 months), and it seems to be an LDC specific issue (so not a language thing).
See commit-per-commit for an explanation of the whats and whys.