[PATCH] libc++: Make support for thread-unsafe C functions optional

Jonathan Roelofs jonathan at codesourcery.com
Mon Mar 30 15:43:11 PDT 2015



On 3/30/15 4:38 PM, Marshall Clow wrote:
> On Mon, Mar 30, 2015 at 2:46 PM, Eric Fiselier <eric at efcs.ca
> <mailto:eric at efcs.ca>> wrote:
>
>     In http://reviews.llvm.org/D8703#149021, @ed wrote:
>
>     > Inhttp://reviews.llvm.org/D8703#148896, @EricWF wrote:
>     >
>     > > The CMake and LIT stuff LGTM. However I would rather keep the directory structure in test/std the same as the structure of the C++ standard clauses. Could you rename the fail tests, put them in the correct directories and use `// REQUIRES: libcpp-has-no-thread-unsafe-c-functions`?
>     >
>     >
>     > Sure! Just to make sure I understand it correctly: in this specific case, the tests should be called:
>     >
>     >   test/std/utilities/date.time/asctime.fail.cpp
>     >   test/std/utilities/date.time/ctime.fail.cpp
>     >   test/std/utilities/date.time/gmtime.fail.cpp
>     >   test/std/utilities/date.time/localtime.fail.cpp
>     >
>     >
>     > As in, the same as they are right now, but without the no.thread.unsafe.c.functions component in between?
>
>
>     Pretty much, but I would make the test names more descriptive so
>     they infer that these are failure tests for when we have no unsafe C
>     functions. The test names don't *have* to follow the standard names.
>
> The rule is that each (leaf) folder corresponds to a section in the
> standard.
> Since there is no section named "[no.thread.unsafe.c.functions]", we
> shouldn't have a folder with that name.

Sorry, my bad for suggesting that.

Jon

> -- Marshall
>

-- 
Jon Roelofs
jonathan at codesourcery.com
CodeSourcery / Mentor Embedded



More information about the cfe-commits mailing list