-
-
Notifications
You must be signed in to change notification settings - Fork 150
feat: add prod and dev attributes to npm_link_targets
#2051
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
feat: add prod and dev attributes to npm_link_targets
#2051
Conversation
|
e0a193d to
5ef2ba5
Compare
| """\ | ||
| # buildifier: disable=function-docstring | ||
| def npm_link_all_packages(name = "node_modules", imported_links = []): | ||
| def npm_link_all_packages(name = "node_modules", imported_links = [], prod = False, dev = False): |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Any other ideas for this API?
It might be nice if we didn't have this scenario where 1/4 of the states is basically invalid and has an odd non-intuitive default. Maybe just throw when false/false and then the other 3 states are valid and understandable?
dev = True|False|None or something like that is one alternative, although the word "dev" in the API without "prod" in the API is confusing imo. Or type = "dev"|"prod"|None but I'd want a term better then "type".
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Actually is this API simply a copy of npm_translate_lock? In which case we should do exactly as you have and my suggestion can be considered in rules_js v3 or something like that...
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Yeah, it's just following the prod/dev API from npm_translate_lock. I agree that an API that excludes invalid states by design would be better for v3.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This current API does not allow a single BUILD to have a mix of these though, correct? If you wanted to include dev deps for some targets (such as tests) but not for others (such as libraries).
Would something such as :node_modules[__{dev,prod}] as the API also solve this?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
It does allow a mix. You could write npm_link_targets() and npm_link_targets(prod = True) and npm_link_targets(dev = True) in the same BUILD if you wanted.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I was thinking you can't have multiple npm_link_all_packages() targets though...?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Do we need the npm_link_all_packages API changes in this PR at all?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I can't remember why I included the npm_link_all_packages changes originally; I guess just for parallelism with npm_link_targets. We're only using npm_link_targets(prod = True).
| if bazel_package == "js/private/worker/src": | ||
| link_2(name = "{}/abortcontroller-polyfill".format(name)) | ||
| link_targets.append("//{}:{}/abortcontroller-polyfill".format(bazel_package, name)) | ||
| link_targets.append("//{}:{}/abortcontroller-polyfill".format(bazel_package, name)) if (not prod and not dev) or (prod and not True) or (dev and True) else None |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
This generated if ... is ugly and I'm pondering how else we can do it. I think we can generate something simpler.
Can't we just compute a few vars at the top of the method such as include_dev and include_prod? Then this will be if include_dev else None or if include_prod else None or blank (or if True else None instead of blank).
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Yeah, something like that should be possible I think.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
After taking another look, I don't think it is possible. The contents of this fragment depend on whether the package in question is a production or dev dependency for the bazel_package in question. I.e., it's not a simple include_dev / include_prod boolean for each package, but rather a matrix of bazel_package x link_alias, which we're encoding as a nested conditional.
There might be some other way to structure this, but it would require a more invasive change.
|
IIUC this solves the issue where rules_js currently treats the lockfile |
This is my suggested solution for #1879. I think it's orthogonal to the lockfile |
npm/private/transitive_closure.bzl
Outdated
| "dev_deps": dev_deps, | ||
| # all deps of this importer to link in the node_modules folder of that Bazel package and | ||
| # make available to all build targets; this includes devDependencies | ||
| "all_deps": all_deps, |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Is this still used?
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
No, will remove.
|
@jfirebaugh have you had a chance to make any updates here? Are you using the patch in your repo? WDYT of
Or would that add more confusion with things like Then |
ede5b45 to
90378de
Compare
|
Still using this in our repo. Just updated the patch for 2.3.8 and removed the change to |
90378de to
ede5b45
Compare
|
Hmm, removing that breaks something for us. Will have to investigate later. |
|
@jfirebaugh this has been picked up in #2298, would you be able to test that out to see if it works for you? |
|
#2298 has merged, let us know if that is missing anything |


Fixes #1879
Changes are visible to end-users: yes
Test plan