At 07:45 AM 10/28/2002, Greg Stein wrote: >On Mon, Oct 28, 2002 at 07:01:15AM -0500, Jeff Trawick wrote: >>... >> We need matching apr and apr-util trees which are just as stable.
++1 :-) >> That isn't necessarily what apr needs as it gears up for 1.0. > >Note that APR has already decided on a versioning scheme (insofar as any >decision is firm for longer than a few months :-). Correct... APR is well on its way to releasing 1.0, in fact, IIRC the big showstopper is finishing the versioning API and some stand-alone build details. >HTTPD should pick a version of APR and stick with that. In particular, it >should stop trying to include the latest CVS into its source tree. As >APR(UTIL) has stabilized, there is less and less of a reason for HTTPD to >track HEAD. Of course, if HTTPD requires a change in APR(UTIL), then the >right people can apply the patch, crank out a new dot release, and then you >bounce the min requirement for APR(UTIL) in the HTTPD bits. If we maintain backwards compat, HTTPD_2_0_RELEASE will always hang out back in APR_0_9_final. httpd can continue to trust the APR api through the end of this cycle. APR 1.0 flushes away all the deprecated APIs. This is why I suspect that httpd 2.2.0 release would coincide with the APR 1.0 release. The library project has rules defined on forwards and backwards compatibility once the APR 1.0 point occurs. Bill
