Olivier PRENANT <ohp(at)pyrenet(dot)fr> writes:
> I just tris again and 3 of the test still failed (with 7.2b1)
That looks good: the diffs from the reference files are exactly the
PST-vs-PDT-in-1947 diffs that are expected on SCO-derived platforms.
I've committed these as abstime-solaris-1947.out etc.
I'd expect that you are still seeing some failures in the timetz test,
too --- but that seems to be an oversight in the regression test itself,
not a platform-specific difference. I assume Thomas will fix the test
shortly...
regards, tom lane