Remember failures from previous builds #937
Conversation
|
@chalin This should at least provide all the functionality you need in order to make sure builds fail always with --fail-on-severe, even if its not entirely clear exactly what caused the failure in the followup build case. |
| /// | ||
| /// See [markActionFailed] and [markActionSucceeded] for more info. | ||
| final Map<int, Set<AssetId>> _failedActions; | ||
| UnmodifiableMapView<int, Iterable<AssetId>> get failedActions => |
There was a problem hiding this comment.
why do we need to expose this as UnmodifiableMapView instead of Map?
There was a problem hiding this comment.
Just being defensive, I want people to go through the methods to update this because its important to remove the keys when the set is empty (so that you can use failedActions.isNotEmpty to see if there were any failed actions).
There was a problem hiding this comment.
I wasn't clear...
It makes sense to me to wrap it in UnmodifiableMapView for defensiveness.. I'm not sure why the getter's signature also needs to say that.
There was a problem hiding this comment.
Oh it probably doesn't I guess, it makes it a bit more clear that its not modifiable though?
There was a problem hiding this comment.
The interface is the same, the methods you can't use will still autocomplete in your editor etc. It's only if you hover over it and look for the type that you'd notice.
There was a problem hiding this comment.
It's only if you hover over it and look for the type that you'd notice.
Ya but that is a fairly common use case right? At least that is usually how I figure out the type of a thing.
Why wouldn't we want to expose this type?
There was a problem hiding this comment.
🤷♂️
I'm fine with it if you think it adds value.
Further work towards #910, we will now remember all failures (regardless of --fail-on-severe flag), and if you provide that flag we will fail the build.
The current message is somewhat lacking, looks roughly like this:
I added another action item to #910 to add a
--rerun-failing-actionsflag to get more information. I don't think its worth trying to store all the information about previous failures as it could get quite large if one failure caused a large number of other cascading failures.