[fdo-internals] For review: 64-bit linux compilation support

[Jackie Ng]

Ok, what was mostly derailing the unit tests was the bad calls to realpath()
in 64-bit Linux.

So I was able to run the tests, but there were segfaults in various parts of
the FDO core and some providers in 64-bit Linux, and some test failures due
to data truncation. I'll put up updated logs for these failures against that
same ticket when I have time.

Just to make sure these changes have not affected our existing builds, I did
a (compilation/test) of this sandbox against windows and 32-bit Linux. And
there were test failures there too, but these test failures *also* existed
in a build of the previous stable branch (3.7). So either my environment is
a bit out of wack, or failing tests are genuine failures and haven't been
fixed for a while.

Either way, my sandbox is producing the same test results as the older 3.7
branch. Where there's failures, the same failures happen in 3.7. No new test
failures have been introduced.

I've aborted my attempts at full 64-bit Linux support (due to my aggressive
release schedule for MapGuide 2.5), but this set of changes will at least
let you compile the full FDO core and providers (ArcSDE unknown) on 64-bit
Linux. The actual segfaulting and portability problems can be addressed for
a future release (3.9?)

Here's the revised unified set of changes:

http://trac.osgeo.org/fdo/changeset?new=6772%40sandbox%2Flinux64&old=6693%40trunk

- Jackie



--
View this message in context: http://osgeo-org.1560.n6.nabble.com/For-review-64-bit-linux-compilation-support-tp5029140p5032951.html
Sent from the FDO Internals mailing list archive at Nabble.com.