-
Notifications
You must be signed in to change notification settings - Fork 5.4k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Add support for [dev-dependencies]
to forc
#3265
Comments
OK, this should be good to go! ## Highlights - Adds support for multiple entry points to IR. Adds a tests for serializing entry functions to/from IR. - Enables ASM generation for libraries (to support test function entry points). - Track entry points through ASM generation so we can return entry point metadata as a part of the compilation result. This doesn't affect ASM generation, but allows tools using `sway-core` as a library to work with different entry points. - Updated E2E test harness with a new test category "UnitTestsPass". - Added E2E tests with multiple unit tests for each type of Sway program (library, script, predicate, contract). - Gets `forc test` working with pretty output! Here's the output from the new `lib_multi_test`: ![Screenshot from 2022-11-03 19-00-57](https://user-images.githubusercontent.com/4587373/199700362-ba32f90d-1f0f-4f76-bf89-b191de9589af.png) ## Test running implementation Currently, it seems like there's no publicly accessible approach to execute a script directly from a custom entry point. As a result, for each test, we patch the bytecode with a `JI` instruction that jumps from after the data section setup to the test's entry point. This is a bit hairy, but works for now! As a follow-up, we may want to consider adding support upstream in `fuel-vm` for executing scripts directly from a given entry point. Even if this was only exposed from the interpreter API, this would make the `forc-test` implementation quite a bit cleaner and avoid the need to patch the bytecode. Alternatively, when building projects we could return more metadata along with the compiled output (whether in memory or bytecode header) to indicate how to work with different entry points in a more reliable manner (rather than the magic const offset currently used in `forc-test`). # TODO - [x] Add `include_tests` flag to `BuildConfig`, allowing `forc` to trigger compilation of test functions. - [x] Include `#[test]` fns as entry points within dead code analysis. - [x] IR and ASM generation for test entry points. - [x] Add `forc build --tests`. - [x] Add `forc test`. - [x] Always include test fns in `TyProgram` (for DCA), but omit from IR if not building tests. - [x] Work out how to iterate over different entry points during `forc test`. - [x] Only generate tests for top-level "members" (not all dependencies). ### Follow-up: - #3260 - #3261 - #3262 - #3263 - #3264 - #3265 - #3266 - #3267 - #3268 Closes #1832.
I can see that tackling this one will help with #3264 too. Since we may end up introducing this After we have this tackled I feel like #3264 will be only:
|
If we think about it more, we may want to introduce concept of Then we can make use that feature to add these I am in between implementing this as a custom case or like this, as a more generalized solution + set of nice defaults , any input is appreciated 🙌 |
Eventually we'll want to allow people to specify dependencies that are only introduced when building tests or other "dev"-only builds (e.g. benchmarks).
See cargo's docs on their equivalent feature: https://doc.rust-lang.org/cargo/reference/specifying-dependencies.html#development-dependencies
Dev dependencies will likely always need to be included in the
Forc.lock
file, but will be omitted from the actual build process unless building for aforc test
invocation.The text was updated successfully, but these errors were encountered: