r239846 - parser: diagnose empty attribute blocks

[Aaron Ballman]

On Wed, Jun 17, 2015 at 12:00 AM, Saleem Abdulrasool
<compnerd at compnerd.org> wrote:
> On Tue, Jun 16, 2015 at 1:21 PM, Aaron Ballman <aaron at aaronballman.com>
> wrote:
>>
>> On Tue, Jun 16, 2015 at 4:18 PM, Meador Inge <meadori at gmail.com> wrote:
>> > On Tue, Jun 16, 2015 at 3:12 PM, Aaron Ballman <aaron at aaronballman.com>
>> > wrote:
>> >
>> >> I would prefer this to be worded differently. It's specific to
>> >> Microsoft's attributes, and "attribute block" isn't a term of art I'm
>> >> used to. How about:
>> >
>> > FWIW, it is the same error that the VC++ compiler emits:
>> >
>> > https://msdn.microsoft.com/en-us/library/0ka5hw13.aspx
>>
>> While reasonable to model after their wording, I still don't think
>> it's a particularly good way to word the diagnostic, especially since
>> these can be used in conjunction with C++-style attributes. We could
>> use "Microsoft attribute block cannot be empty" if "block" is a
>> sensible term that others want.
>
>
> Im not particularly tied to the message itself.  At the end of the day, the
> messages should be useful to users, not the author of the support.  Changed
> to your suggestion in SVN r239879.

Thank you! Change looks good.

~Aaron