Conversation
|
So I currently expect this to fail, since compatibility with iOS was broken somewhere in this diff d54c78a...b51f25e which we discovered when updating envoy and envoy-mobile's usage of the rules. My assumption is that it was caused by c41020e which always assumes |
|
cc @jsharpe from the commit above hoping you have some ideas here |
|
You can see the current failure here https://buildkite.com/bazel/rules-foreign-cc/builds/3452#a07e12a5-3042-484f-98d2-c60cdbe57a95 |
8ea64a8 to
ba50ced
Compare
|
Ok confirmed that was the issue and pushed a fix here. I was thrown off while trying to reduce because of #769 |
a90b8b6 to
0ac298b
Compare
|
I don't see how this could have affected the Linux python3 setup, but I will try to debug more |
Yes, that's an odd failure; I can't see how this change would have affected the RBE python3 build either. |
Some projects like envoy-mobile use rules_foreign_cc to build targeting iOS, so this is useful to validate on CI as part of the macOS test build.
0ac298b to
db9b684
Compare
|
My current assumption is that by changing the workspaces here I'm triggering a build that was previously cached, and somehow rotted, maybe through a docker image update |
62b4a25 to
dcae5f5
Compare
|
Ok I take it back, it seems to be a workspace ordering issue. I think because of some transitive dependency on |
jsharpe
left a comment
There was a problem hiding this comment.
Thanks for getting to the bottom of this! It's very useful to have more diverse use cases in the tests
|
❤️ thanks! |
Some projects like envoy-mobile use rules_foreign_cc to build targeting
iOS, so this is useful to validate on CI as part of the macOS test
build. This also fixes a bug where this was broken since
c41020e because the wrong SDKROOT was used