In the reviewing system, after a patch is approved, what should be the following step?
I assume I’ll have to make sure the patch hasn’t broken anything, before pushing it, so I’ll wan’t to run a full build+tests, probably in a remote sterile environment (like a jenkins server).
Do we have an integration system like this? Or should I just trust the build+tests on my machine?
Regardless, I have seen in many reviews (in Phabricator) that Builds are triggered for them, in particular: “arc lint + arc unit” and “pre-merge checks” (which sounds like what I am searching for). What is their use and how do I trigger them?
the pre-merge tests are in prototype state. Once they are ready for testing, I’ll announce it on the mailing list
Unless things have changed lately and I missed something, you can just push the change and the bots will pick it up automatically. They will send you emails if it seems like anything is broken.
That being said, before pushing you should do some local tests (e.g. run the lit tests) as an initial sanity check.
If it breaks anything, the standard process is
- Revert it with an explanation of why in the commit message. (E.g. “This broke this bot” with a link to the build log or a simple example.)
- Fix the breakage
Thanks. I suspected it to be this way.
Thanks for the update.
What about “arc lint + arc unit”?