> Things are indeed very stable now, at least on 32-bit machines. But
> 64-bit machines are catching on, along with Firebird going to 2.x,
> both of which will probably entail changes in python and kinterbasdb.
There was a trivial preprocessor-related problem that prevented
kinterbasdb 3.1 from compiling out of the box with Firebird 2.0a1, but
that's fixed in both CVS HEAD and the maint-3-1 branch.
I tested kinterbasdb on 64-bit Linux (Fedora Core/AMD64) with
prereleases of Firebird 1.5.1 about a year ago
(http://sourceforge.net/mailarchive/message.php?msg_id=8356329 ), and
have continued to do so since. There were two or three 64-bit
cleanness issues in kinterbasdb back then, which I fixed.
A more likely source of problems (though not on AMD64) is
big-endianness. I haven't so much as compiled kinterbasdb, or any
other software, on a big-endian machine. Even though I have an
adequate understanding of endianness, it'd be easy to let a bug of that
sort slip by at 02:15 some morning. I suppose I could use the SF
compile farm for that purpose, but I haven't gotten around to it.
Wednesday, May 04, 2005
kinterbasdb (python) on 64bit
David Rushby responded about 64bit issue