[svn.haxx.se] · SVN Dev · SVN Users · SVN Org · TSVN Dev · TSVN Users · Subclipse Dev · Subclipse Users · this month's index

Re: svn commit: r40377 - trunk/subversion/libsvn_ra_serf

From: Paul Burba <ptburba_at_gmail.com>
Date: Fri, 6 Nov 2009 09:31:36 -0500

On Thu, Nov 5, 2009 at 2:22 PM, Daniel Shahaf <d.s_at_daniel.shahaf.name> wrote:
> Paul T. Burba wrote on Wed, 4 Nov 2009 at 19:26 -0800:
>> Author: pburba
>> Date: Wed Nov  4 19:26:27 2009
>> New Revision: 40377
>>
>> Log:
>> Follow-up to r38105, fix some diabolical ra_serf failures in the merge tests.
>>
>> * subversion/libsvn_ra_serf/util.c
>>   (svn_ra_serf__handle_xml_parser): Do not depend on order of evaluation
>>   of the arguments in the call to svn_error_compose_create() since this is
>>   not defined by the C standard
>
> This bitten us a few times already.

Hi Daniel,

I hadn't realized that, do you recall where (this is really just for
my own curiosity)?

> Perhaps we could add a macro that
> forces the right order?  Or enable a compiler warning for this?
>
> e.g.,
>
> #define svn_error_compose_create2(errvar1, errvar2, expr1, expr2) \
>    (errvar1 = (expr1), errvar2 = (expr2),                        \
>     svn_error_compose_create(errvar1, errvar2))
>
> #define svn_error_compose_create2(errvar, expr1, expr2)    \
>  do {                                                     \
>    svn_error_t __err1 = (expr1);                          \
>    svn_error_t __err2 = (expr2);                          \
>    errvar = svn_error_compose_create(__err1, __err2);     \
>  } while (0)

Maybe...but isn't any argument list to any function that depends on
the evaluation order of those arguments just as likely to be a
problem?

Paul

------------------------------------------------------
http://subversion.tigris.org/ds/viewMessage.do?dsForumId=462&dsMessageId=2415106
Received on 2009-11-06 15:32:11 CET

This is an archived mail posted to the Subversion Dev mailing list.