With this change, the parse step is bypassed for MacOS. The bnf grammar was updated to test the PR.
Unfortunately, this won't fix the spurious hangs of the MacOS runners that we see. Google has a problem with the runner code and/or is using too few servers, or is using the slowest servers on the planet. However, bypassing the MacOS tester is a small price to pay to increase the chance for the tests to be completed in a reasonable amount of time. We need a compilation and link for C++ under the MacOS environment because it is the rate-limiting factor in the code we write for the C++ target.
Just yesterday, in antlr/antlr4, a PR was submitted in which 39 runners completed in a minute or two on Windows and Ubuntu. However, the PR sat around for a MacOS test to complete in 1h 25m, with the setup of PHP taking the majority of the time at 1h 23m!
teverett
commented
5 days ago
This PR fixes #4326.
With this change, the parse step is bypassed for MacOS. The bnf grammar was updated to test the PR.
Unfortunately, this won't fix the spurious hangs of the MacOS runners that we see. Google has a problem with the runner code and/or is using too few servers, or is using the slowest servers on the planet. However, bypassing the MacOS tester is a small price to pay to increase the chance for the tests to be completed in a reasonable amount of time. We need a compilation and link for C++ under the MacOS environment because it is the rate-limiting factor in the code we write for the C++ target.
Just yesterday, in antlr/antlr4, a PR was submitted in which 39 runners completed in a minute or two on Windows and Ubuntu. However, the PR sat around for a MacOS test to complete in 1h 25m, with the setup of PHP taking the majority of the time at 1h 23m!