I'm closing in on a v2.2 release. Here's the current list of things
remaining, all of which are polishing matters, and (hopefully) small bug
fixes:
o We probably don't need both platform.h and defs.h. Fold them
together and rework the rest of the library to cope.
o Test that it still works under MinGW after the recent
DLL linkage fixes for VC++.
o RPM build shouldn't require Bakefile. This may mean building
examples even though they're not needed. One way around
this may be if we can say something like "make libmysqlpp.so"
in the build process instead of "make all".
o Move to a single Bakefile probably breaks tarball build
and more. Also requires changes to README type files.
o Can simplify Windows Bakefile coverage in HACKERS file
by just giving the "bakefile -f msvc6prj mysql++.bkl"
command. Then you don't need Cygwin or bakefilize.
o Add function to library to return its version number in a
structured form. Then, make resetdb query it to ensure that
it's linked to the same version. Prevents errors caused
by not using exrun script when running examples with a
different library version installed.
o resetdb is hanging in Query::execute() on RHEL 3 test system.
Find out why.
The most important thing that needs testing is the VC++ support. A few
days ago, I think I solved the build problem. VC++ building got broken
the last time we fixed MinGW. As a consequence, it'd also be nice to
get reports of MinGW testing, to see whether we've re-broken it.
Beware that the library and examples are no longer built separately in
the VC++ case. There is now just a single top-level .dsw file, as this
makes dependency checking easier, and prevents confusion about what to
build, when. This was done separately from the DLL linkage fix; it's
just been one of those things I wanted to get done, so I did it while I
was already working on the Windows side. I only bring it up because
it's changed since the last time I asked anyone to test the VC++
version, so if you check it out on top of an existing copy and don't
realize that this changed, your build may break for reasons I would find
uninteresting. :)
Those of you who are watching development may notice that some features
were dropped from this release in the interest of getting the thing out
the door. The BLOB/binary data stuff is still top of my wishlist for
the next release; I've just decided that I need to release this thing
soon, if only because people need some of the bug fixes in it.
In unrelated news, I hereby announce that this release of MySQL++ is
codenamed Longhorn.
