Uh oh!
There was an error while loading. Please reload this page.
gh-122931 CI for Linux multiarch co-installability - #152831
Conversation
a347830 to
3a52cb5Compare3a52cb5 to
b84e719Comparestefanor
commented
Jul 1, 2026
Duh. Let me store hash manifests instead of storing the installs. |
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
webknjaz
commented
Jul 2, 2026
Do you mean the debug option?
Seems like having another matrix will add 15–16 minutes of CI runtime. |
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
Uh oh!
There was an error while loading. Please reload this page.
| required: false | ||
| type: string | ||
| default: '' | ||
| upload-install-hashes: |
There was a problem hiding this comment.
Looks like this input is always passed exactly opposite to bolt-optimizations. A case could be made that it could be computed instead...
There was a problem hiding this comment.
They may currently be opposite, but I think they are orthogonal, I wouldn't want to compute one from the other.
upload-install-hashes should be set once for each (architecture, build flags) combination.
I could imagine adding a check that aborts a build if upload-install-hashes is enable and any other non-architcture, non-build-flag option is non-default. Would that improve things?
| if: inputs.upload-install-hashes | ||
| uses: actions/upload-artifact@bbbca2ddaa5d8feaa63e36b76fdaad77386f024f # v7.0.0 | ||
| with: | ||
| name: ${{ env.INSTALL_HASHES_FILE }} |
There was a problem hiding this comment.
Should this also expose a wildcard for the consumers to look up the artifacts? Though, this probably won't work with a matrix 🤔
Uh oh!
There was an error while loading. Please reload this page.
It isn't helping make anything more reproducible, setting it so late. We'll file a bug about `.pyc` reproducibility, later.
As suggested in this thread, add some CI tooling to ensure that Python continues to be multi-arch co-installable.
Explicitly skip checking
abi3.soandabi3t.soextensions, as they are currently not co-installable (#122931).We extend
reusable-ubuntuto perform an extra set of build+install+hash runs. This required some slightly different configure options to get more co-install-friendly builds. Those configure options broke some existing tests, but the fixes are trivial, and included.From an earlier implementation that didn't use hash-manifests:
Example failure (without
abi3andabi3tignored): https://github.com/stefanor/cpython/actions/runs/28550736484/job/84647793000Example success (with them ignored): https://github.com/stefanor/cpython/actions/runs/28551134448/job/84648971317