On Mon, 2002-11-04 at 16:29, Brad Settlemyer wrote: > Could you please explain some of the benefits of encoding the version number > into the namespace name? > > I'm worried that this is going to cause me some of the same problems that > having the version name encoded into the library causes.
Major "API versions" absolutely should be in the library name. How else will an application benefit from bugfixes in the Xerces-C++ library that is uses, without relinking that application against a new version of the library. This is all quite normal. Apparently this is fixed now (I can't tell because _no_ library is installed by "make install" now), but in the past it was very annoying to have to release new versions of applications just so that they worked with slightly-updated newer versions of Xerces. > I know that > several integrators have complained about that on this list multiple times, > and that many of the developers, myself included, are building the library > to the name libxerces.xxx manually to avoid these problems. I'm worried > that this is just placing a xerces internal implementation detail into my > software, and I can't figure out what the extra value is. The proposal did seem to say that a non-versioned namespace name could be used to refer to the latest version, so I don't think this would be visible to you unless you want to use that functionality. -- Murray Cumming [EMAIL PROTECTED] www.murrayc.com --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
