Graham Reitz wrote:
> I made a little bit of progress. I was statically linking the
> libmysqlclient_r.a and dynamically linking the mysqlpp library. When I
> dynamically linked both the memory alignment message is no longer present.
> Although, I don't understand why it should matter.
Lo, a crack of light doth shine 'pon the wreckage. Perhaps you built
MySQL++ against the C API dynamic library, then linked your test program
against the static version? This would result in two copies of all
MySQL routines in the final binary, and more importantly, a mixture of
data structures from one or the other, depending on whether they're
created by code within MySQL++ or whether they don't get instantiated
until your test program is built.
The more I think about it, the more I like this hypothesis, because I
don't see how you can link MySQL++ against a static MySQL library
without changing mysql++.bkl. You're just not given the option to do it
any other way.
Other than that, I don't see why mixing linkage types should matter. As
long as you're consistent in what type of each library is linked
throughout the project, it shouldn't care. This is GCC after all.
> I should probably go ahead and build Fink,
It's not as hard or time-consuming as you might think.
One of my biggest complaints about Fink is that it seems to build a lot
of stuff from source even if you tell it to prefer binaries, so going
with the purely source-bootstrapped version seems like a difference
without a distinction.