[clang] [clang-tools-extra] [clang-tidy] Add new check `modernize-use-designated-initializers` (PR #80541)

[Sam McCall via cfe-commits]

Danny =?utf-8?q?Mösch?= <danny.moesch at icloud.com>,
Danny =?utf-8?q?Mösch?= <danny.moesch at icloud.com>,
Danny =?utf-8?q?Mösch?= <danny.moesch at icloud.com>,
Danny =?utf-8?q?Mösch?= <danny.moesch at icloud.com>,
Danny =?utf-8?q?Mösch?= <danny.moesch at icloud.com>,
Danny =?utf-8?q?Mösch?= <danny.moesch at icloud.com>,
Danny =?utf-8?q?Mösch?= <danny.moesch at icloud.com>,
Danny =?utf-8?q?Mösch?= <danny.moesch at icloud.com>,
Danny =?utf-8?q?Mösch?= <danny.moesch at icloud.com>,
Danny =?utf-8?q?Mösch?= <danny.moesch at icloud.com>,
Danny =?utf-8?q?Mösch?= <danny.moesch at icloud.com>,
Danny =?utf-8?q?Mösch?= <danny.moesch at icloud.com>,
Danny =?utf-8?q?Mösch?= <danny.moesch at icloud.com>,
Danny =?utf-8?q?Mösch?= <danny.moesch at icloud.com>,
Danny =?utf-8?q?Mösch?= <danny.moesch at icloud.com>,
Danny =?utf-8?q?Mösch?= <danny.moesch at icloud.com>,
Danny =?utf-8?q?Mösch?= <danny.moesch at icloud.com>,
Danny =?utf-8?q?Mösch?= <danny.moesch at icloud.com>,
Danny =?utf-8?q?Mösch?= <danny.moesch at icloud.com>,
Danny =?utf-8?q?Mösch?= <danny.moesch at icloud.com>,
Danny =?utf-8?q?Mösch?= <danny.moesch at icloud.com>,
Danny =?utf-8?q?Mösch?= <danny.moesch at icloud.com>,
Danny =?utf-8?q?Mösch?= <danny.moesch at icloud.com>,
Danny =?utf-8?q?Mösch?= <danny.moesch at icloud.com>,
Danny =?utf-8?q?Mösch?= <danny.moesch at icloud.com>,
Danny =?utf-8?q?Mösch?= <danny.moesch at icloud.com>,
Danny =?utf-8?q?Mösch?= <danny.moesch at icloud.com>,
Danny =?utf-8?q?Mösch?= <danny.moesch at icloud.com>,
Danny =?utf-8?q?Mösch?= <danny.moesch at icloud.com>,
Danny =?utf-8?q?Mösch?= <danny.moesch at icloud.com>,
Danny =?utf-8?q?Mösch?= <danny.moesch at icloud.com>,
Danny =?utf-8?q?Mösch?= <danny.moesch at icloud.com>,
Danny =?utf-8?q?Mösch?= <danny.moesch at icloud.com>,
Danny =?utf-8?q?Mösch?= <danny.moesch at icloud.com>,
Danny =?utf-8?q?Mösch?= <danny.moesch at icloud.com>,
Danny =?utf-8?q?Mösch?= <danny.moesch at icloud.com>,
Danny =?utf-8?q?Mösch?= <danny.moesch at icloud.com>,
Danny =?utf-8?q?Mösch?= <danny.moesch at icloud.com>,
Danny =?utf-8?q?Mösch?= <danny.moesch at icloud.com>,
Danny =?utf-8?q?Mösch?= <danny.moesch at icloud.com>,
Danny =?utf-8?q?Mösch?= <danny.moesch at icloud.com>,
Danny =?utf-8?q?Mösch?= <danny.moesch at icloud.com>,
Danny =?utf-8?q?Mösch?= <danny.moesch at icloud.com>,
Danny =?utf-8?q?Mösch?= <danny.moesch at icloud.com>,
Danny =?utf-8?q?Mösch?= <danny.moesch at icloud.com>,
Danny =?utf-8?q?Mösch?= <danny.moesch at icloud.com>,
Danny =?utf-8?q?Mösch?= <danny.moesch at icloud.com>,
Danny =?utf-8?q?Mösch?= <danny.moesch at icloud.com>,
Danny =?utf-8?q?Mösch?= <danny.moesch at icloud.com>,
Danny =?utf-8?q?Mösch?= <danny.moesch at icloud.com>,
Danny =?utf-8?q?Mösch?= <danny.moesch at icloud.com>,
Danny =?utf-8?q?Mösch?= <danny.moesch at icloud.com>,
Danny =?utf-8?q?Mösch?= <danny.moesch at icloud.com>,
Danny =?utf-8?q?Mösch?= <danny.moesch at icloud.com>,
Danny =?utf-8?q?Mösch?= <danny.moesch at icloud.com>,
Danny =?utf-8?q?Mösch?= <danny.moesch at icloud.com>,
Danny =?utf-8?q?Mösch?= <danny.moesch at icloud.com>,
Danny =?utf-8?q?Mösch?= <danny.moesch at icloud.com>,
Danny =?utf-8?q?Mösch?= <danny.moesch at icloud.com>,
Danny =?utf-8?q?Mösch?= <danny.moesch at icloud.com>,
Danny =?utf-8?q?Mösch?= <danny.moesch at icloud.com>,
Danny =?utf-8?q?Mösch?= <danny.moesch at icloud.com>,
Danny =?utf-8?q?Mösch?= <danny.moesch at icloud.com>,
Danny =?utf-8?q?Mösch?= <danny.moesch at icloud.com>
Message-ID:
In-Reply-To: <llvm.org/llvm/llvm-project/pull/80541 at github.com>


sam-mccall wrote:

TL;DR: sounds like I should revert the removals/deps in clangd, and let you decide how to structure clang-tidy's copy of the code?

`tidy::utils` as a dep has a conceptual layering problem, and a few practical ones: it brings in a pile of dependencies that aren't otherwise in the no-tidy build (effect: much longer build times), and it sets the wrong expectations around performance: that meeting the batch-latency requirements of clang-tidy is sufficient, rather than the interactive-latency requirements of clangd. (Not a theoretical concern: clangd's *deliberate* deps on clang-tidy have had multiple unacceptable performance regressions in the past, which is the cost of taking the dep, but not one that needs paying here).

> This were main reason, why it were avoided, as there were no unit tests for this code on clangd side.

This wasn't a library in its own right on the clangd side. If it's a library with multiple clients, it needs tests so problems with the library can be distinguished from problems with the clients. (I think it would have been nice to have fine-grained unittests anyway, but we didn't).

> Why is it such a big deal that a revert needs to be considered?

It's not a big deal, but I thought the most likely fix was to move this into a separate library with tests, I wasn't sure if anyone wanted to take that on in a hurry, and wanted to get back to a good state.

If tidy owners are happier with just cloning the code, I can revert just the changes to clangd.

I would suggest moving the code out out tidy/utils and into the check if that's the only place it's going to be tested, but that's up to you.

https://github.com/llvm/llvm-project/pull/80541