diff options
author | Alex Eagle <eagle@post.harvard.edu> | 2020-10-26 07:11:16 -0700 |
---|---|---|
committer | GitHub <noreply@github.com> | 2020-10-26 07:11:16 -0700 |
commit | 5be1f76e3ecd1f743f4213f3087c2a0961411782 (patch) | |
tree | 3fe17d491344694ab562a571bd0a797dd302fc64 /README.md | |
parent | 69f55c10e9a77e334800e6266ab43be0e320fa30 (diff) | |
download | bazelbuild-rules_python-5be1f76e3ecd1f743f4213f3087c2a0961411782.tar.gz |
Remove py_repositories (#370)
This adds useless setup code to a users WORKSPACE file.
The idea that we could add some transitive WORKSPACE dependencies is flawed without a working support for this in Bazel. Anything we add in that function later is a breaking change for users, in that they might call our py_repositories() before fetching rules_xx, and we already installed an incompatible rules_xx.
Since adding anything here is a breaking change, we can always put it back later as a breaking change to the rules.
However I would argue strongly that rules_python is too core in the dependency chain for it to *ever* grow transitive dependencies. Like rules_nodejs we should either vendor code we need privately or strip development-only dependencies from our distribution, and never suggest that users call a rules_python function that installs other starlark code.
Diffstat (limited to 'README.md')
-rw-r--r-- | README.md | 3 |
1 files changed, 0 insertions, 3 deletions
@@ -86,9 +86,6 @@ http_archive( strip_prefix = "rules_python-{}".format(rules_python_version), url = "https://github.com/bazelbuild/rules_python/archive/{}.zip".format(rules_python_version), ) - -# Then load and call py_repositories() as -# above. ``` Once you've imported the rule set into your `WORKSPACE` using any of these |