Cache dependencies irrespective of test failures.#450
Conversation
|
One of the things to test/be careful for is if the dependencies/build somehow gets corrupted, causing the build failures. I haven't tested this personally, but I think you could run into a situation where you have a new cache, something fails the build, and then you store that corrupted/failed build dependency into the cache. |
|
@joshmgross Yeah, we’re encountering that unhappy eventuality in this patch already 😭 |
|
Also looks like none of the post cache actions ran, there might be a bug with how we treat |
|
Doesn’t look like it wrote to the cache after failing to build dependencies, FWIW? |
|
The cache action defines a Turns out this will override anything defined in the main |
Guessing that’s not something we can really do anything about from this end, then. At the very least, it doesn’t appear that the |
|
@robrix Yep, sorry about that, I thought this scenario would work. You could fork the action and modify that Feel free to file an issue on https://github.com/actions/cache for full feature support |
This PR should allow us to cache updates to dependencies even in the presence of test failures, reducing iteration time.
🎩 @joshmgross for the suggestion.