This example could be used by projects that build Rust
end product crates on an OS matrix, but share the cargo cache.
To avoid fetching the dependencies from the network in
each matrix job in case of a cache miss, a dedicated job
pre-populates the cache on Linux, to be reused by the
jobs in the matrix.
Also demonstrate separate proximate caching of the registry index.
As the packed crate sources are stored under ~/.cargo/registry/cache,
it's redundant to archive the unpacked sources.
The sufficient local state of the git dependencies is in
~/.cargo/git/db. Cargo automatically checks out the source
trees if missing.
* Allow for multiple line-delimited paths to cache
* Add initial minimatch support
* Use @actions/glob for pattern matching
* Cache multiple entries using --files-from tar input
remove known failing test
Quote tar paths
Add salt to test cache
Try reading input files from manifest
bump salt
Run test on macos
more testing
Run caching tests on 3 platforms
Run tests on self-hosted
Apparently cant reference hosted runners by name
Bump salt
wait for some time after save
more timing out
smarter waiting
Cache in tmp dir that won't be deleted
Use child_process instead of actions/exec
Revert tempDir hack
bump salt
more logging
More console logging
Use filepath to with cacheHttpClient
Test cache restoration
Revert temp dir hack
debug logging
clean up cache.yml testing
Bump salt
change debug output
build actions
* unit test coverage for caching multiple dirs
* Ensure there's a locateable test folder at homedir
* Clean up code
* Version cache with all inputs
* Unit test getCacheVersion
* Include keys in getCacheEntry request
* Clean import orders
* Use fs promises in actionUtils tests
* Update import order for to fix linter errors
* Fix remaining linter error
* Remove platform-specific test code
* Add lerna example for caching multiple dirs
* Lerna example updated to v2
Co-Authored-By: Josh Gross <joshmgross@github.com>
Co-authored-by: Josh Gross <joshmgross@github.com>
Tested this with my own repo which uses a mix of `build.gradle` and `build.gradle.kts` files and this glob seems to be working correctly.
As an aside, please checkout #215 as it would make the process of verifying these globs easier!
* Adding examples for esy as a workflow for OCaml files
* track v1 instead of v1.1.2
Co-Authored-By: Josh Gross <joshmgross@github.com>
* add link in the readme for ocaml-esy
* ocaml -> ocaml/reason
* link in readme says ocaml/reason
Co-authored-by: Josh Gross <joshmgross@github.com>
* Use different IDs for 1) getting the directory of yarn cache 2) the cache itself
Using the current example + https://github.com/actions/cache#skipping-steps-based-on-cache-hit,
I came to a wrong conclusion that I could skip a step
if the `cache-hit` was `true` -
the ID I used was from the wrong step -
the `get yarn cache directory` step,
instead of the `get yarn cache itself` step.
I've updated the example in hopes that it'll be clearer for others aswell!
Signed-off-by: Kipras Melnikovas <kipras@kipras.org>
* Explain which ID to use for `cache-hit` in yarn's example
Signed-off-by: Kipras Melnikovas <kipras@kipras.org>
