I think the point was to put one simple number in the bakefile, and in
turn, it will put 8 tons of crap in autoconf for you. I just installed
bakefile, and played around a little but I'm having a set of completely
new and unrelated problems that's preventing me from seeing it come to
fruition. Bakefile 0.2.2 vs. your 0.2.0 seems to add a lot more stuff
to Makefile.in, including some stuff about BK_DEPS and AROPTIONS that
isn't working for me. On the plus side, it does appear to have added a
bunch of stuff to Makefile.in, and doesn't seem to make a dent in the
mingw stuff etc.
adding this to mysql++.bkl
<libname>mysqlpp</libname>
<so_version>2.2.1</so_version>
and doing a rebake gets me a lot of new stuff about so_name in
Makefile.in, for different operating systems. It seems to require
another configure.ac macro though, and I got very confused when trying
to find out what to add, for another test. running "autoconf" on
unmodified source configure.ac got me similar errors though.
-Jack
Warren Young wrote:
> Remi Collet wrote:
>>
>> Using -soname, the .so (link) must be provided by mysql++-devel
>> sub-package as it is not required to run binaries.
>
> That's sensible. Thanks for clearing that up for me.
>
>> I've look at bakefile documentation (quite poor on this feature).
>
> I've contributed several patches to the Bakefile docs, but yes, it still
> needs work.
>
>> There
>> are options for "dll" (dllname, libname, version and so_version) but i
>> don't know how to use them...). Need more search/try.
>
> Again, I don't think Bakefile is the right place for this. Bakefile is
> platform-independent, whereas the change you're talking about is not.
> The right place for it is as an autoconf test. Then you just substitute
> the flags into the generated Makefiles the same way we do for pthreads
> and probably other things I'm forgetting right now.
>
