You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
In #898, Windows CI was enabled in 2.1, but it ignores prebuilt baseline failures. It makes sense that Windows is different enough that it pulls in different prebuilts.
I'm not aware of any scenario where prebuilts for the Windows build matter, but filing this to track.
(Prebuilt baseline failure might happen to detect when we use a prebuilt package rather than a source-built package, but cache conflict detection catches this situation in a much more targeted way. It would have a lot more value to fix what it detects regularly (even baseline it?) and change that into an error.)
The text was updated successfully, but these errors were encountered:
A situation this can catch is if a submodule builds the wrong packages and downstream repos pull in prebuilts instead. Wrong packages can be built e.g. if the portable build passing code breaks due to a submodule update: CoreCLR builds portable, then Core-Setup builds non-portable and tries to find non-portable CoreCLR bits.
We have a separate sanity check to detect when no packages are built, but it doesn't catch wrong packages.
In #898, Windows CI was enabled in 2.1, but it ignores prebuilt baseline failures. It makes sense that Windows is different enough that it pulls in different prebuilts.
I'm not aware of any scenario where prebuilts for the Windows build matter, but filing this to track.
(Prebuilt baseline failure might happen to detect when we use a prebuilt package rather than a source-built package, but cache conflict detection catches this situation in a much more targeted way. It would have a lot more value to fix what it detects regularly (even baseline it?) and change that into an error.)
The text was updated successfully, but these errors were encountered: