-----BEGIN PGP SIGNED MESSAGE-----
On 25 Mar 2001, at 16:07, Tom Lane wrote:
> Does that database have any user-created relations in it, or is it
> just a virgin database? It seems that the wrong attlen is being
> computed for ctid fields during bootstrap, but the regression test
> output (if it was complete) implies that the value inserted for
> user-created fields was OK. This doesn't make a lot of sense since
> it's the same code...
Totally virgin. I created it just for that select you wanted. The
7.1beta6 I built was installed in /usr/pgsql so as to be entirely
separate from any other running parts of the system. Like I said, the
test failed, but it seems to *work* just fine...
If you want the complete regress output, I'll send it as well. The
only failures were the type_sanity and geometry though, and the
geometry was just fluctuations on the final digit of a few numbers.
I suspect it might be an alignment problem (ARM needs word or dword
alignment on data access.. our kernel has an alignment trap handler
that does fixups in 'broken' code) or something related to signedness
(ARM has default signed char) but I don't know enough about postgres
internals to really debug it. However, I'm certainly willing to
learn.. :)
-----BEGIN PGP SIGNATURE-----
Version: N/A
iQCVAwUBOsAFXf+IdJuhyV9xAQFpZgP8C7g9dqlh9Qd/wVwJn2jquVh+X3gBWBZ5
UMHx43tPfYE7xJvHl3XH/z+mg/POyzgFMCF+5USO2jzbPMDiS2OtJbp+1NvP2FHA
uuY1ra5o8WKWW/7ZrfaO5edC5e1OsKbhGsXugRIyBwFkzz28blt6gongUdio0nC3
Td8Fm3GUKNk=
=+//W
-----END PGP SIGNATURE-----