| FazBrowse GitHub Viewer | Trending | | Home |
| Tools: [Download Repo ZIP] [Original HTTPS Page] |
| "macos-14", # M1 | ||
| "macos-13", # Intel | ||
| ] | ||
| os: ${{fromJson(inputs.os-matrix)}} |
There was a problem hiding this comment.
We can't pass arrays into reusable workflows, only strings, booleans or numbers. So we pass the array as a string, then use fromJson to convert the string into a list.
Sorry, something went wrong.
Sorry, something went wrong.
Only test free-threading with faster macOS M1
Only test free-threading with faster macOS M1
Only test free-threading with faster macOS M1
| Back | FazBrowse Home | New Git URL |
We used to only have a single macOS Intel job in the CI.
But we have recently quadrupled it:
Plus we occasionally have a couple testing the JIT, but they only trigger when relevant files are modified.
Unfortunately, we're limited to a maximum of 5 concurrent macOS jobs:
This means at busy times, jobs can be queued for a long time waiting for free macOS runners. In the meantime, the other jobs have finished, but we need to sit around until one of the five runners becomes free.
We've run in to this once per week for the last three weeks, during European evening/US daytime.
To mitigate, let's remove some macOS jobs.
We can't remove them from the regular build, because they're needed to maintain PEP 11's Tier 1 support. So let's start by removing one macOS test job for the free-threading build.
Here's macOS job times for five recent builds. macos-13 is Intel, macOS-14 is M1.
M1 is much faster (~8 mins), so let's remove the slower free-threading Intel job (~20 mins).
If the problem persists, we can remove the free-threading M1 job as well.
And hopefully GitHub will be able to increase the number of concurrent macOS jobs in the future, especially as the free-threading work matures. 🤞