Summary: By getting rid of callbacks, we get a code that's pretty halved in size, and much easier to read, but we also fix bugs! When some of the code of Module would throw, we wouldn't catch it, meaning that packager would crash entirely. That's the case for when we have a Haste name mistach, for example. This changeste fixes that by instead capturing exceptions through the Promise chain, that bubble up all the way to the `Server` and reasonnably handled. People can then fix the Haste nane mismatch and refresh right away, and it'll work as expected.
Reviewed By: davidaurelio
Differential Revision: D5137408
fbshipit-source-id: b9e938b93e8d23738af49106bbae1fce97d7a5db
Summary:
This changeset moves the creation of the transform cache at the top-level of the bundler so that:
* we can use alternative folders, such as the project path itself, that I think will be more robust especially for OSS;
* we can disable the cache completely, that is useful in some cases (for example, the script that fills the global cache).
The reasons I believe a local project path is more robust are:
* there are less likely conflicts between different users and different projects on a single machine;
* the cache is de facto cleaned up if you clone a fresh copy of a project, something I think is desirable;
* some people have been reporting that `tmpDir` just returns nothing;
* finally, it prevents another user from writing malicious transformed code in the cache into the shared temp dir—only people with write access to the project have write access to the cache, that is consistent.
Reviewed By: davidaurelio
Differential Revision: D5113121
fbshipit-source-id: 74392733a0be306a7119516d7905fc43cd8c778e
Summary: This is a first step towards doing an experiment I was talking about, that is, using the local directory (the directory where the config file lives) instead of the tmp dir, that is fragile.
Reviewed By: davidaurelio
Differential Revision: D5112326
fbshipit-source-id: 819636209972115e41213867f3eb78fbdf6ed9df
Summary:
Changes the global cache to use *local paths* rather than base names of files for the cache key. This is to enable correct usage of babel transforms like `transform-react-jsx-source` that use file paths in their output.
It remains a responsibility of the transform implementer to pass relative paths to babel if the global cache is being used.
Reviewed By: jeanlauliac
Differential Revision: D5044028
fbshipit-source-id: 2ef1e1545e510a18ab49a307053d91b4f40269b2
Summary: Internally when adding `format` we have a lint rule that automatically reformat using `prettier` and the project rule. I'm concerned how we can ensure this stays consistent when merging PRs though. It's not a big issue because the volume of PRs is low, but we'll have to figure it out later on.
Reviewed By: davidaurelio
Differential Revision: D5035830
fbshipit-source-id: 6f2bc9eb8212938ff785a34d2684efd1a9813e1a
Summary: This removes the single-use "garbage collection" class and make a single `TransformCache` instead. The reason for doing this is that I'd like to experiment with using a local dir based on the root path instead of the `tmpdir()` folder, because on some platform using the temp dir is subject to permissions problem, sometimes it is empty, and on all platform it is vulnerable to concurrency issues.
Reviewed By: davidaurelio
Differential Revision: D5027591
fbshipit-source-id: e1176e0e88111116256f7b2b173a0b36837a887d
Summary: deduplicates / unifies types for source maps across the code base
Reviewed By: jeanlauliac
Differential Revision: D4955924
fbshipit-source-id: 25cb71031dce835dd7d2bc1c27d6b20050906e81
Summary: It seems we don't need that one much anymore, as we have a FS-based cache for files, and removing the last callsites doesn't really change perf, as we have to read these files eventually (plus some of these are read from `HasteMap`, that I believe is already dead code).
Reviewed By: davidaurelio
Differential Revision: D4884220
fbshipit-source-id: 4cf59f16a6f0bdf275abe81e9de2f34816866bae
Summary:
Note: if this changeset causes some breakage, consider disabling rather than reverting. To disable, the call to `_preprocessPotentialDependencies` in `ResolutionRequest` can be removed.
It's a bit of an experiment. I couldn't see any particular regression caused by this, but I could see net improvement of the global cache performance, as it unlock much, much stronger batching: indeed, instead of discovering dependencies progressively, we synchronously figure out the list of all potential modules of a bundle, and kick-off cache fetching and/or transformations. So when it comes to fetching from the global cache, it'll do less requests, and each requests will ask for considerably more keys at a time.
Potential problem caused by this changeset: if a module's dependencies completely changed, then the first time we try to build the bundle it'll start transforming modules that we probably don't care at all anymore, spending precious CPU time for nothing. I've been thinking about it and I cannot see such a case happening much often. Even if it happens, it should not cause any bug or corruption, it would just take additional time.
Other potential problem: that this new code doesn't handle some types of edge cases. It's quite hard to figure out what could possibly break in the `ResolutionRequest` code (and I think it would benefit from a larger refactor). We do have a good test coverage for `DependencyGraph` and it seems to work smoothly, so I'm relatively confident we're not breaking edge cases.
Reviewed By: davidaurelio
Differential Revision: D4875467
fbshipit-source-id: 2dfcc755bec638d3d1c47862ec1de5220953e812
Summary: The original code wasn't very readable. Better with this change, I think.
Reviewed By: davidaurelio
Differential Revision: D4851335
fbshipit-source-id: 210309d4b727aff58bea48d0ab324256234cd941
Summary:
One of my changeset broke the "ModuleGraph" code without warning earlier because we are using `any`, that equivalent to having no typing at all. This changeset fixes the types so that `ResolutionRequest` is exactly what it actually is: a class usable for any `Module`-looking class, including the normal one, and the "ModuleGraph" one used for Buck builds. That way, the ModuleGraph's `Module` is typechecked against `Moduleish`.
Concretely this change mostly migrates the `Module` to its generic parameter counterpart `TModule` inside `ResolutionRequest`.
Reviewed By: kentaromiura
Differential Revision: D4826256
fbshipit-source-id: fcd7ca08ac6c35e4e9ca983e2aab260e352bcb4e
Summary: We want any exception thrown by `_getHasteName` to be captured by the promise instead of breaking the outer stack.
Reviewed By: davidaurelio
Differential Revision: D4754825
fbshipit-source-id: 173c7c8867da73efb198ed3159704d6fd0e7b87d
Summary:
Similar to https://github.com/facebook/jest/pull/2877, this introduces an optional config `HasteImpl` of type `{getHasteName(filePath: string): (string|void)}` that returns the haste name for a module at filePath if it is a haste module or undefined otherwise.
This allows us to inject a custom implementation of haste's module id resolution rather than only relying on `providesModule` annotations
Reviewed By: davidaurelio
Differential Revision: D4589372
fbshipit-source-id: 4d1983dfbf09c9d67faf725e86ae86ab42433b7d
Summary:
This moves the `src` directory one level up and removes the `react-packager` folder. Personally, I always disliked this indirection. I'm reorganizing some things in RNP, so this seems to make sense.
Not sure if I forgot to update any paths. Can anyone advice if there are more places that need change?
Reviewed By: jeanlauliac
Differential Revision: D4487867
fbshipit-source-id: d63f9c79d6238300df9632d2e6a4e6a4196d5ccb
