[PATCH] D93844: [clang-format] Add possibility to be based on parent directory
Zhao Wei Liew via Phabricator via cfe-commits
cfe-commits at lists.llvm.org
Mon Dec 27 04:25:52 PST 2021
zwliew added inline comments.
================
Comment at: clang/lib/Format/Format.cpp:3067
+ if (!ChildFormatTextToApply.empty()) {
+ assert(ChildFormatTextToApply.size() == 1);
+
----------------
HazardyKnusperkeks wrote:
> zwliew wrote:
> > Is there a reason behind limiting the number of children configs to 1 in this case? When the fallback is not used, more than 1 children configs are allowed (line 3036).
> >
> > Sorry for digging this up, I came across this seemingly arbitrary decision when working on some changes to https://reviews.llvm.org/D72326
> Yeah but it doesn't happen, there is at most only one `.clang-format` in the parent directory path which is found. So we assert on that and then don't need to loop over what is exactly one element big.
Thanks for the reply. However, I don't think that's true. Yes, it's only possible to find one `.clang-format` file in the first parent directory. But if it has `BasedOnStyle: InheritParentConfig` set, then we could find another `.clang-format` file in the "grandparent" directory. In this case, we'll have 2 elements in `ChildFormatTextToApply`, but only the very first one will actually be applied.
To illustrate, suppose we have the following file structure:
```
- .clang-format
- foo
|- .clang-format
|- input.cpp
```
Both `.clang-format` files have `BasedOnStyle: InheritParentConfig` set. When running `clang-format --style=file foo/input.cpp`, only the inner config is applied on the fallback style, while the outer config is ignored. When testing on a debug build, I encountered a crash due to the failed assert. When removing the assert, and using a loop to apply the configs, both the inner and outer configs are applied, which I believe is the expected behaviour.
Repository:
rG LLVM Github Monorepo
CHANGES SINCE LAST ACTION
https://reviews.llvm.org/D93844/new/
https://reviews.llvm.org/D93844
More information about the cfe-commits
mailing list