Hey,
I would like to ask if it is worth a shot to create a pull request which basically clang-formats projects like libcxxabi/libunwind ?
Since those projects aren't following the LLVM coding standards.
It would make the code even "more readable" also it would give "me" the feeling that it actually belongs to the LLVM project.
Thanks
-- Peet
Hi Peet,
We had this discussion a number of times for different projects and
the general feeling was that it wouldn't be a big problem to do so,
but that also depends on the project's community. I'm copying some of
the main developers of those projects for their view on the matter.
I personally prefer not to format code that hasn't changed, as that
messes up with the version history (git blame and all), but people are
normally less paranoid than I am.
cheers,
--renato
I personally prefer not to format code that hasn't changed, as that
messes up with the version history (git blame and all), but people are
normally less paranoid than I am.
+1
> I would like to ask if it is worth a shot to create a pull request which
basically clang-formats projects like libcxxabi/libunwind ?
> Since those projects aren't following the LLVM coding standards.
> It would make the code even "more readable" also it would give "me" the
feeling that it actually belongs to the LLVM project.Hi Peet,
We had this discussion a number of times for different projects and
the general feeling was that it wouldn't be a big problem to do so,
but that also depends on the project's community. I'm copying some of
the main developers of those projects for their view on the matter.I personally prefer not to format code that hasn't changed, as that
messes up with the version history (git blame and all), but people are
normally less paranoid than I am.
Normally, I would agree. However, unless Im mistaken, the splitting of
libunwind did destroy history :-(. In that case, unless there are
objections from others (I think Nick should be included in this set), I
don't think we loose too much in the specific case of libunwind. But, I
agree in the case of libc++/libc++abi.
There’s probably a way to abuse the -textconv argument to git (diff|blame)…