Tatsuo Ishii <t-ishii@sra.co.jp> writes:
> Ok. I have tested patches from Bruce. Now tests for bit passed.
> Remaining issues seem that strtol() is broken on SunOS4, not detecting
> an overflow, which causes int4 and some of other tests failure.
Looks like you might also want to use horology-no-DST-before-1970.out.
> Should we use our own strtol()?
Mumble. The memcmp fix was all from well-tested parts, but I don't
think we have a canned test for strtol breakage available. Also,
I believe that the SunOS strtol lossage has been known and tolerated
in previous Postgres releases. Since it's not a regression, I'm going
to change sides and vote with Peter: let's not fix this one now.
regards, tom lane