Re: clar unit testing framework FTBFS on uclibc systems (wchar_t unsupported)
- From
Patrick Steinhardt <ps@pks.im>
- Date
- Oct 21, 2024, 06:44 UTC
- Message-ID
- <ZxX4YtZ5lW0axToT@pks.im>
- In-Reply-To
- <ZxJnfYtuxnAEBc1E@archie.me>
On Fri, Oct 18, 2024 at 08:49:49PM +0700, Bagas Sanjaya wrote:
> On Thu, Oct 17, 2024 at 03:54:28PM +0200, Patrick Steinhardt wrote:
[snip]
> On Buildroot site, Edgar Bonet (Cc:'ed) suggests to improve your patch by > wrapping strcmps [1]:
Thanks. So with the proposed patch things work?
One thing I still don't understand: why don't you get an error from includeng <wchar.h>? As shown above, it should contain the following snippet:
> > #ifndef __UCLIBC_HAS_WCHAR__ > > #error Attempted to include wchar.h when uClibc built without wide char support. > > #endif
So I'd expect that to trigger and cause the build to abort. Am I missing anything here? Let me have another look at uclibc...
Oh, yes! There's also a "wchar-stub.h" file that replaces "wchar.h" when compiled without UCLIBC_HAS_WCHAR=Yes. And that file indeed does not cause us to error out, but is a stub. It also explains why the logs do not complain about the missing `wchar_t` type, as that type does get defined by the stub header. Weird, but so be it.
I've picked this up upstream via [1].
[1]: https://github.com/clar-test/clar/pull/108
Patrick