> On 2011-03-22 12:14:43, Andrew Stitcher wrote: > > Looking at the gcc documentation, it seems that the way > > __attribute((visibility=..)) is supposed to be used is the opposite way > > round to which we use in our headers (and in Visual Studio/Windows) your > > supposed to set a default which is more visible and then override it with > > less visible member functions. > > > > Doing it that way looks like it would avoid the warnings you are seeing (I > > haven't tried this though so I may have it wrong in practice) > > Alan Conway wrote: > I didn't get the same impression, I'll read again. The big advantage in > doing it this way is that windows does it this way and the windows folks have > already defined and tested our public API for us with macros we can re-use. I > also feel like it makes more sense to me to directly define the public API > rather than indirectly defining it as the inverse of the non-public > functions. Seems like it is more error prone to try to mark every non-public > function - since that's a constantly changing set - than mark the stable set > of public API functions.
I completely agree with this, just stating my impression of the gcc documentation - Andrew ----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/520/#review347 ----------------------------------------------------------- On 2011-03-22 08:10:48, Alan Conway wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > https://reviews.apache.org/r/520/ > ----------------------------------------------------------- > > (Updated 2011-03-22 08:10:48) > > > Review request for qpid and Andrew Stitcher. > > > Summary > ------- > > WIP: enable -fvisibility=hidden -fvisibility-inlines-hidden > > Failing with warnings: warning: lowering visibility of ... to match its type > Google suggests warning is ok, but no apparent way to disable it. > Try build with warnings off to see if the results are good. > > > Diffs > ----- > > trunk/qpid/cpp/configure.ac 1084053 > trunk/qpid/cpp/include/qmf/engine/QmfEngineImportExport.h 1084053 > trunk/qpid/cpp/include/qpid/CommonImportExport.h 1084053 > trunk/qpid/cpp/include/qpid/ImportExportDefs.h PRE-CREATION > trunk/qpid/cpp/include/qpid/agent/QmfAgentImportExport.h 1084053 > trunk/qpid/cpp/include/qpid/client/ClientImportExport.h 1084053 > trunk/qpid/cpp/include/qpid/console/ConsoleImportExport.h 1084053 > trunk/qpid/cpp/include/qpid/messaging/ImportExport.h 1084053 > trunk/qpid/cpp/src/Makefile.am 1084053 > trunk/qpid/cpp/src/qpid/broker/BrokerImportExport.h 1084053 > > Diff: https://reviews.apache.org/r/520/diff > > > Testing > ------- > > > Thanks, > > Alan > >
