[PATCH] Disallow clang from assigning a diagnostic to ID zero

[David Majnemer]

Updated version of the patch, just made it a little cleaner.


CC'ing Argyrios Kyrtzidis as he, from svn history, seems to know the area.


On Tue, Jul 2, 2013 at 2:57 PM, Richard Smith <richard at metafoo.co.uk> wrote:

> LGTM, but I'd like someone who knows this area better than I do to look at
> it.
>
> --- lib/Basic/DiagnosticIDs.cpp (revision 184534)
> +++ lib/Basic/DiagnosticIDs.cpp (working copy)
> @@ -82,11 +84,9 @@
>  #include "clang/Basic/DiagnosticSemaKinds.inc"
>  #include "clang/Basic/DiagnosticAnalysisKinds.inc"
>  #undef DIAG
> -  { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0}
>  };
>
> Are you sure nothing is using this to find the end of the table?
>
> On Tue, Jul 2, 2013 at 2:45 PM, David Majnemer <david.majnemer at gmail.com>
> wrote:
> > Ping
> >
> >
> > On Fri, Jun 21, 2013 at 3:42 AM, David Majnemer <
> david.majnemer at gmail.com>
> > wrote:
> >>
> >> The fallout of r184401 made it apparent that clang uses diagnostic ID
> zero
> >> as a sentinel value while *also* assigning it to an actual diagnostic,
> >> err_attribute_not_type_attr.
> >>
> >> The attached patch should do two things:
> >> * Ensure that we don't give away ID 0
> >> * Make sure that we will assert if we attempt to use ID 0
> >>
> >> --
> >> David Majnemer
> >
> >
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/cfe-commits/attachments/20130717/c5cc6673/attachment.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: diag0.diff
Type: application/octet-stream
Size: 3063 bytes
Desc: not available
URL: <http://lists.llvm.org/pipermail/cfe-commits/attachments/20130717/c5cc6673/attachment.obj>