Oh, and if you respond with "but the build keeps breaking because
people leave this in", then I have a very simple response: don't paper
over the problem. The problem is that developers were NEVER supposed
to leave them in. Teach the developer to be more careful. Don't give
them a crutch.
On Fri, May 20, 2011 at 07:38, Greg Stein <gstein_at_gmail.com> wrote:
> No.
>
> By design, we did not want to leave these scattered around the code
> base. SVN_DBG() is generally for personal use only (people want to
> print different things at different times, for different reasons). The
> entire *idea* was to make it create a build failure for non-DEBUG
> situations.
>
> I very much disagree with this change.
>
> On Fri, May 20, 2011 at 06:44, <stefan2_at_apache.org> wrote:
>> Author: stefan2
>> Date: Fri May 20 10:44:31 2011
>> New Revision: 1125316
>>
>> URL: http://svn.apache.org/viewvc?rev=1125316&view=rev
>> Log:
>> Fix build breakage: teach the preprocessor to simply ignore SVN_DBG
>> in non-debug builds.
>>
>> * subversion/include/private/svn_debug.h
>> (SVN_DBG): define as no-op, if SVN_DEBUG has not been defined
>>
>> Modified:
>> subversion/trunk/subversion/include/private/svn_debug.h
>>
>> Modified: subversion/trunk/subversion/include/private/svn_debug.h
>> URL: http://svn.apache.org/viewvc/subversion/trunk/subversion/include/private/svn_debug.h?rev=1125316&r1=1125315&r2=1125316&view=diff
>> ==============================================================================
>> --- subversion/trunk/subversion/include/private/svn_debug.h (original)
>> +++ subversion/trunk/subversion/include/private/svn_debug.h Fri May 20 10:44:31 2011
>> @@ -85,5 +85,9 @@ svn_dbg__printf(const char *fmt, ...)
>> }
>> #endif /* __cplusplus */
>>
>> +#else /* SVN_DEBUG */
>> +
>> +#define SVN_DBG(ARGS) (void)
>> +
>> #endif /* SVN_DEBUG */
>> #endif /* SVN_DEBUG_H */
>>
>>
>>
>
Received on 2011-05-20 13:40:21 CEST