Conversation
8W9aG
commented
Nov 27, 2023
- Local always results in true.
- Unclear why the flag is useful in this case.
- Currently impedes builds using bazel RBEs.
* Local always results in true. * Unclear why the flag is useful in this case. * Currently impedes builds using bazel RBEs.
|
This looks like it undoes #29, which had quite a bit of discussion. I expect that merging this PR will break what was fixed by #29. Tagging everybody who commented on #29: Could you please advise? — @quval, @SomeoneSerge, @TheButlah, @jordanzliu, @dmadisetti, @dstripelis, @QuantamHD, @chuckx
I believe that would be much safer. |
|
If I remember correctly, what #29 actually did (among the rest) was adding the option to disable local execution. I'm not using |
|
I am fine with doing either (removing or bringing the option up to the main rule). If I could put my "what to do" hat on for a second though, I think its a bit strange there is a local force and we don't know why it exists. It seems to me like this is a leftover from a debugging session that somehow made its way into production code. I think it might be better to remove it and wait for something to break, and if so re-enable it, and leave a comment as to why this exists. I suppose I think that even if we go with the propagation method, people might be confused when this rule doesn't work for RBEs, and its not immediately obvious that they will need to manually switch a flag on the rule (they will assume this is done by the bazel settings implicitly). Having said that, it does introduce a risk for the maintainers as potentially breaking a production component, so I understand if that doesn't want to be taken. Happy to go with either direction. |
OK, sounds good. Let's just remove it for now. If it's needed by anyone later, maybe it's better to propagate From the docs:
|
|
For the record: testing with pybind11_abseil and pybind11_protobuf passes. |