[llvm] r216170 - Make format_object_base's destructor protected and non-virtual.

[Chandler Carruth]

What Patrik said. =D This LGTM once we're confident that for ToT Clang, we
enable the warning. I like the approach taken, it seems much nicer than a
version check!


On Sat, Aug 23, 2014 at 1:06 PM, Patrik Hägglund H <
patrik.h.hagglund at ericsson.com> wrote:

> Hi Dylan,
>
> I was about to commit the version check Chandler proposed. But, a small
> test case seems even better. However, for clang-3.5rc2 I still get
>
> -- Performing Test CXX_WONT_WARN_ON_FINAL_NONVIRTUALDTOR - Failed
>
> Perhaps you need to add -std=c++11 to the check?
>
> When fixed, please commit!
>
> /Patrik Hägglund
>
> -----Original Message-----
> From: dylan.noblesmith1 at maine.edu [mailto:dylan.noblesmith1 at maine.edu] On
> Behalf Of Dylan Noblesmith
> Sent: den 23 augusti 2014 20:37
> To: Chandler Carruth
> Cc: Patrik Hägglund H; llvm-commits at cs.uiuc.edu
> Subject: Re: [llvm] r216170 - Make format_object_base's destructor
> protected and non-virtual.
>
> On Sat, Aug 23, 2014 at 6:09 AM, Chandler Carruth <chandlerc at google.com>
> wrote:
> >
> > On Sat, Aug 23, 2014 at 2:58 AM, Patrik Hägglund H
> > <patrik.h.hagglund at ericsson.com> wrote:
> >>
> >> > In all seriousness, we need to find a solution to this soon, or revert
> >> > this change.
> >>
> >>
> >>
> >> +1
> >>
> >>
> >>
> >> > For those who want to use old/pre-installed/official release clangs
> >> > rather than more recent iterations, they can just disable –Werror
> (it's not
> >> > the default anyway).
> >>
> >>
> >>
> >> For us, disabling –Werror is currently not an option. If this is the
> >> intention, I think the policy should be updated first. (I would rather
> like
> >> the policy to be strengthened, to support –Werror builds with the latest
> >> official release of _both_ clang and gcc.)
> >
> >
> > While I haven't really had time, and I didn't even write the original
> patch,
> > I've already described exactly how to fix this by disabling the warning
> with
> > a version check in CMake... It should be really straightforward to
> > implement. I guess if for some reason none have done it I will, but I
> don't
> > even have a version of Clang to test it with, so I'm somewhat inclined
> for
> > someone who does to write this patch...
> >
> Patch attached, to disable the warning when it gives false positives
> like that. OK to commit?
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-commits/attachments/20140823/662f1788/attachment.html>