[cfe-dev] [clang] How to ignore compiler warnings from macros in system headers?

[Carlos Galvez via cfe-dev]

Thanks for the response! I went ahead and put up a patch
<https://reviews.llvm.org/D116833> to introduce support for disabling
warnings from macros on a per-warning level. It's implemented following the
great feedback I received and works for my concrete use case (disable a
particular warning). After it's merged people should be able to suppress
other warnings if they see fit. However, as I said in the comments, I
believe this behavior should be consistent with "ShowInSystemHeader", i.e.
explicitly opt-in to get warnings from macros instead of opt-out. But
that's not up to me to decide, I'm keeping current behavior (all warnings
on by default) :)

/Carlos



On Mon, Jan 10, 2022 at 8:23 PM David Blaikie <dblaikie at gmail.com> wrote:

> +Some folks who have some interest in clang diagnostic quality.
>
> I'm guessing there isn't an existing solution to this in clang - but maybe
> it's possible if you wanted to work on an improvement/patch to Clang?
> Though it might be difficult to check that all the relevant code properties
> came from the macro and not some from the macro, some from macro arguments,
> some from the context the macro was used in, etc.
>
> On Sun, Dec 26, 2021 at 2:29 PM Carlos Galvez via cfe-dev <
> cfe-dev at lists.llvm.org> wrote:
>
>> Hi!
>>
>> I've asked this question in StackOverflow
>> <https://stackoverflow.com/questions/70488470/how-to-ignore-compiler-warnings-for-macros-defined-in-third-party-headers>
>> without much success so I thought I'd come closer to the source and ask
>> here.
>>
>> My use case is simple: ignore compiler warnings from 3rd-party headers
>> that I have no control over. The standard approach for this is simple -
>> include them as "system headers" via "-isystem". That works pretty well,
>> except for one bit - *macros*. Clang (and GCC) will still report
>> warnings if I use problematic macros each time I use them in my code. Why
>> is that? Clang knows where the macro comes from, shouldn't it be able to
>> ignore the warning? Is there a good way to achieve what I want, without
>> using pragmas or re-creating the third-party code myself?
>>
>> Thanks!
>> /Carlos
>> _______________________________________________
>> cfe-dev mailing list
>> cfe-dev at lists.llvm.org
>> https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-dev
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/cfe-dev/attachments/20220110/9273bfff/attachment-0001.html>