Skip to content

Conversation

EliahKagan
Copy link
Owner

@EliahKagan EliahKagan commented Aug 2, 2025

The purpose of this fork-internal PR is to facilitate a squash while being able to link to details of a testing methodology and associated CI results, where including the individual steps as separate commits (or even describing each of them individually in the commit message) are nonetheless not really justified.

This relates to GitoxideLabs#2093 shall be merged as a squash commit into a feature branch, either for that PR or a successor PR.


As noted in GitoxideLabs#2092, the wasm jobs on CI do not test gix-diff directly. However, my prediction there that breakage would not be detected on CI was mistaken, because those jobs do test the wasm feature of gix-pack. The gix-pack crate depends on gix-diff, and its wasm feature enable the gix-diff one.

This nonetheless adds an explicit check for gix-diff. This gix-diff check does not attempt to build default features, since some fail on some WASM targets. But the preexisting gix-pack check does still build the default features of gix-pack, which are compatible with WASM targets.

The efficacy of these checks, as well as the need to pass --no-default-features for gix-diff, can be confirmed by examining CI results for various fragments of this change experimented on in #75. This also demonstrates that CI is capable of catching at least some breakages related to the wasm feature of gix-diff, and thus may be sufficient to support moving forward with GitoxideLabs#2092.

As noted in GitoxideLabs#2092, the `wasm` jobs on CI do not test `gix-diff`
directly. However, my prediction there that breakage would not be
detected on CI is likely mistaken, because those jobs do test the
`wasm` feature of `gix-pack`. The `gix-pack` crate depends on
`gix-diff`, and its `wasm` feature enable the `gix-diff` one.

This temporary change checks that CI does fail when the `wasm`
feature of `gix-diff` is broken in a simple way, even though it
does not currently run WASM tests of `gix-diff` directly.

This also temporarily makes the `wasm` matrix non-fail-fast, so
that more failures can be observed.
This adds an explicit check of `gix-diff` in the WASM jobs, even
though the `gix-path` check currently covers `gix-diff` (its `wasm`
feature enables the `wasm` feature of its `gix-diff` dependency).
This reverts commit 29bc973, which
temporarily broke the `wasm` feature of `gix-diff` to check CI.
But continue building default features of `gix-pack`.
@EliahKagan EliahKagan changed the title Temporarily break wasm feature of gix-diff to check CI Add gix-diff WASM check, even though gix-path covers it Aug 2, 2025
@EliahKagan
Copy link
Owner Author

Wrapped message text for squash commit:

As noted in #2092, the `wasm` jobs on CI do not test `gix-diff`
directly. However, my prediction there that breakage would not be
detected on CI was mistaken, because those jobs do test the `wasm`
feature of `gix-pack`. The `gix-pack` crate depends on `gix-diff`,
and its `wasm` feature enable the `gix-diff` one.

This nonetheless adds an explicit check for `gix-diff`. This
`gix-diff` check does not attempt to build default features, since
some fail on some WASM targets. But the preexisting `gix-pack`
check does still build the default features of `gix-pack`, which
are compatible with WASM targets.

The efficacy of these checks, as well as the need to pass
`--no-default-features` for `gix-diff`, can be confirmed by
examining CI results for various fragments of this change
experimented on in EliahKagan#75. This also demonstrates that CI is
capable of catching at least some breakages related to the `wasm`
feature of `gix-diff`, and thus may be sufficient to support moving
forward with #2092.

@EliahKagan EliahKagan marked this pull request as ready for review August 2, 2025 22:13
@EliahKagan EliahKagan merged commit 238461a into upgrade-getrandom Aug 2, 2025
43 checks passed
EliahKagan added a commit that referenced this pull request Aug 2, 2025
As noted in GitoxideLabs#2092, the `wasm` jobs on CI do not test `gix-diff`
directly. However, my prediction there that breakage would not be
detected on CI was mistaken, because those jobs do test the `wasm`
feature of `gix-pack`. The `gix-pack` crate depends on `gix-diff`,
and its `wasm` feature enable the `gix-diff` one.

This nonetheless adds an explicit check for `gix-diff`. This
`gix-diff` check does not attempt to build default features, since
some fail on some WASM targets. But the preexisting `gix-pack`
check does still build the default features of `gix-pack`, which
are compatible with WASM targets.

The efficacy of these checks, as well as the need to pass
`--no-default-features` for `gix-diff`, can be confirmed by
examining CI results for various fragments of this change
experimented on in #75. This also demonstrates that CI is
capable of catching at least some breakages related to the `wasm`
feature of `gix-diff`, and thus may be sufficient to support moving
forward with GitoxideLabs#2092.
@EliahKagan EliahKagan deleted the run-ci/upgrade-getrandom-next branch August 2, 2025 22:31
EliahKagan added a commit that referenced this pull request Aug 9, 2025
As noted in GitoxideLabs#2092, the `wasm` jobs on CI do not test `gix-diff`
directly. However, my prediction there that breakage would not be
detected on CI was mistaken, because those jobs do test the `wasm`
feature of `gix-pack`. The `gix-pack` crate depends on `gix-diff`,
and its `wasm` feature enable the `gix-diff` one.

This nonetheless adds an explicit check for `gix-diff`. This
`gix-diff` check does not attempt to build default features, since
some fail on some WASM targets. But the preexisting `gix-pack`
check does still build the default features of `gix-pack`, which
are compatible with WASM targets.

The efficacy of these checks, as well as the need to pass
`--no-default-features` for `gix-diff`, can be confirmed by
examining CI results for various fragments of this change
experimented on in #75. This also demonstrates that CI is
capable of catching at least some breakages related to the `wasm`
feature of `gix-diff`, and thus may be sufficient to support moving
forward with GitoxideLabs#2092.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

Successfully merging this pull request may close these issues.

1 participant