Jon Brandvein

Note that the test you're describing is the subject of FR #16.

Note in particular that admitting arbitrary strings as field names returnable by `getattr`/`dir` impacts serialization, since proto strings must be utf-8.

Hi Hao, thanks for the PR and sorry for the lack of activity. Let me give an update on the state of this repo: We intend to develop a replacement...

I believe this change is small enough that it can bypass the doc review process, if there aren't objections.

Notwithstanding my fixed-width formatting of the term above, "TreeArtifact" isn't actually a separate class, it's represented by `Artifact`. So a separate type might be a little harder to implement due...

Well, naming aside, whatever it is, it seems* ingrained in Bazel that it's to be treated as the same category of thing as an Artifact. \* *(Edit: in my humble,...

Thanks for the info Ulf but I'm a little unclear. I think you're saying these three cases exist: 1. regular Artifact that points to a regular file 2. regular Artifact...

And if it has a trailing '/' character, it would be represented as a TreeArtifact? We just discussed this in the Starlark meeting. The consensus is that we agree in...

We'll leave this issue open a week for potential objections. No design doc because it's a relatively small change. Then we can expose the field and document the subtle issue...

Well if it's intentional that this isn't supported than it certainly should be documented explicitly. I'd argue against allowing bool * string. I believe it's only supported in Python as...