Use standard error format - #154
Conversation
|
@cpovirk The |
|
@netdpb It took me a while to figure out why the conformance tests were not updating for me locally. Is it possible that you need to make a new release of |
|
If you change the samples to use the new format, does the build succeed? If so, I have no objection. On the other hand, I think you'll need to change |
You may have to run with |
I just ran |
Isn't |
|
I've created a new jspecify branch: https://github.com/jspecify/jspecify/tree/samples-google-prototype-eisop |
Both. |
Which task did you run with |
There was a problem hiding this comment.
It would be nice to have a separate (previously merged) PR just to update the conformance test report, so this PR just has difference from that.
There was a problem hiding this comment.
I split that off into #157 and now this PR is just the changes to adapt the error format.
…checker into fix-test
|
Conformance tests and minimal tests pass now. |
In eisop/checker-framework#693 I removed some hacky workaround, not realizing that it is still used here.
As we're not planning to write many such tests, I think it's okay to just use the standard error format here.
Ideally, we would have a mechanism to allow type systems to more easily adapt the expected error lines themselves.
However, at the moment there is a bunch of static methods that make extension hard.