[pacman-dev] Bug in libalpm: sizeof(off_t)

[Jeremy Heiner]

On Wed, Nov 20, 2013 at 10:26 PM, Jeremy Heiner
<scalaprotractor at gmail.com> wrote:
> #if sizeof(off_t) != @???@

Well, that approach seems like it is not going to work. Obviously cpp
can't do sizeof, but I was hoping configure would provide something to
fill that gap. However, now I've read through AC_SYS_LARGEFILE and it
is pretty clear that there's no help there.

Changing alpm.h to use off64_t would be an obstacle to running on
platforms like smartphones, so I assume that is out. There doesn't
seem to be a portable way to keep the compiler from generating .o
files that differ from the libalpm ABI. Nor can the linker detect
those differences (maybe it could be made to, but not without
significant changes to the API).

Leaving only one semi-solution that I can see at this moment: document
this behavior. There's gotta be something I've missed?