On Mon, Jun 04, 2012 at 04:06:20PM -0400, Greg Troxel wrote: > Sandro Santilli <s...@keybit.net> writes: > > > C library clients are fine. C++ library clients are warned. > > Do you mean the "using the C++ library is discouraged" in the docs?
Yes. > That seems to be a hint that it's better to use the C API, not a warning > that the shlib name is unstable. The warning could indeed be extended to mention that. Feel like filing a ticket ? Best if with a patch. > Is there a technical reason why it's good to change the library name > every release? I can see fear about C++ APIs being unstable, but that > means one needs to rebuild when the compiler changes, which is different > From going from 3.3.3 to 3.3.4. It's also ABI being unstable. The technical reason is it being too hard to maintain a stable C++ ABI. See how much it takes for you to tell us if there have been an ABI change between 3.3.3 and 3.3.4. I can't tell. If there's an easy tool to do that I could reconsider, but it would still mean maintaining another set of versions, reaching 3 (one for release, one for C lib, one for C++ lib). > Is this part of a plan to make it difficult to use the C++ library? If > so, I would say that it's not working, because it makes life difficult > for packagers rather than making it difficult for people who release > code that uses the C++ library (postgis and gdal both seem to do that, > or at least they are linked that way on my system - perhaps that's a > configure/packaging bug?). That's a packaging bug. Both PostGIS and GDAL only use the C-API, so they should not be packaged as being dependent on the C++ one. So all clients should depend on libgeos_c, and _only_ libgeos_c should depend on libgeos-XYZ. --strk; ,------o-. | __/ | Delivering high quality PostGIS 2.0 ! | / 2.0 | http://strk.keybit.net - http://vizzuality.com `-o------' _______________________________________________ geos-devel mailing list geos-devel@lists.osgeo.org http://lists.osgeo.org/mailman/listinfo/geos-devel