I think this makes sense, but I'd also like Richard to take a look.

On Tue, Apr 15, 2014 at 6:18 PM, Ben Langmuir <[email protected]> wrote:

> Hi Daniel,
>
> Back in r197485 you made it so that missing module headers are marked
> unavailable and will fail at build/import time rather than when parsing the
> module map file.  This patch fixes the case where a submodule is missing a
> header - right now the top-level module will build without the header,
> which is awful, because 1) you don’t get a diagnostic for the missing
> header and may just get missing symbols, and 2) even after you replace the
> missing header the module won’t rebuild because the pcm file doesn’t depend
> on that header if it wasn’t included.
>
> I’m not sure if I did this the right way, since it seems like the
> MissingHeader should be on the submodule, but I wasn’t sure if we wanted to
> search all of a module’s children in order to figure out what happened.
>
> Ben
>
>
_______________________________________________
cfe-commits mailing list
[email protected]
http://lists.cs.uiuc.edu/mailman/listinfo/cfe-commits

Reply via email to