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
It felt redundant at the time to force enabled: true to turn on a builder, but it turns out that this magic cause some nasty stuff - builders like build_web_compilers:entrypoint start getting turned on for packages other than the root just because someone configured a generate_for or dart2js_args or something.
Changing this would be breaking and it's hard to know how bad the fallout would be.
There isn't currently the capability to express "configure a builder, but only if it would have run anyway."
The text was updated successfully, but these errors were encountered:
In theory we could call this a breaking change on build_config and packages that are doing the "right" think and have a dependency on that package whenever they use build.yaml would be safe - but in practice very few packages are likely doing that.
I think this one is valid as a long-term backlog "next-breaking-release" issue.
I don't think it is worth a breaking change to solve only this issue. If some other use case for a breaking change in the build.yaml format comes up though, I do think it would be worth changing these semantics at the same time.
It felt redundant at the time to force
enabled: true
to turn on a builder, but it turns out that this magic cause some nasty stuff - builders likebuild_web_compilers:entrypoint
start getting turned on for packages other than the root just because someone configured agenerate_for
ordart2js_args
or something.Changing this would be breaking and it's hard to know how bad the fallout would be.
There isn't currently the capability to express "configure a builder, but only if it would have run anyway."
The text was updated successfully, but these errors were encountered: