It's not only proprietary bits for which 32bit compat is useful. There is unfortunately enough
free code still around which is not 64bit clean for various reasons. While it could be cleaned
up it is often a lot of work
and if you just want to get the program to run it is much easier to just
compile/run it as 32bit executable. And it's not that you lose much
from that. After all compat mode is not really much slower.
btw i have a set of shell scripts to make compiling 64bit
as 32bit really easy even with difficult to change Makefiles:
http://firstfloor.org/pub/ak/shell/gccwrap2.tar.gz
Of course you need a proper multilib'ed system.
Posted Apr 15, 2008 13:33 UTC (Tue) by pizza (subscriber, #46)
[Link]
> It's not only proprietary bits for which 32bit compat is useful. There is unfortunately
enough free code still around which is not 64bit clean for various reasons. While it could be
cleaned up it is often a lot of work.
64-bit unclean code tends to happen when the programmer freely interchanged ints and pointers,
or was using on-disk/on-network data structures without fixed-width sizes.
I don't deny that this buggy code exists, but at least in the F/OSS world there's not much of
it left. However, in the "internal company software" world, this is probably more common than
not (going on my personal experience).