[libcxx-commits] [PATCH] D117357: [libc++] Add the "lint" scripts to libcxx-generate-files.

Arthur O'Dwyer via Phabricator via libcxx-commits libcxx-commits at lists.llvm.org
Fri Jan 14 14:09:08 PST 2022


Quuxplusone added a comment.
Herald added a reviewer: ayermolo.

In D117357#3244957 <https://reviews.llvm.org/D117357#3244957>, @ldionne wrote:

> I would like to retain the notion that these are tests, they are not generating anything. How about this:

IIUC, that doesn't solve @Mordante's problem (which also became my problem today, as I realized that I was introducing a new file and didn't want to be arsed about whether `_` sorts before or after `a` — I just wanted to do a quick `ninja some-target` and have it tell me the answer).
Spinning up all of `check-cxx`, or even llvm-lit, not only is wasteful — it adds a //whole separate step// to the overhead of the commit-my-patch workflow. The workflow right now is "edit the files, then if you added any new files `ninja libcxx-generate-files`, then `git commit` the changes." I even remember //me// conservatively objecting to the introduction of `libcxx-generate-files` originally — surely people could just run `libcxx/utils/*.py` by hand! :) But it turns out to be useful. Besides, I personally could remember the path to `../libcxx/util<tabcomplete>/gen<tabcomplete>`, so I didn't see a need for `libcxx-generate-files` originally. But I don't want to have to remember the path to `../libcxx/test/libcxx/lint/<tabcomplete>`!

Alternatively, if you don't like adding the lint steps to `libcxx-generate-files` directly, how would you feel about adding a new cmake target that runs //both// `generate*.py` and `lint*.py`?


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D117357/new/

https://reviews.llvm.org/D117357



More information about the libcxx-commits mailing list