Porting PostgreSQL

Author

HPux 11i

Mark Richardson

2005-07-14, 3:24 am

I've been having a little trouble with compiling on HP. I compiled version 7.4.5 successfully with the command...
make --without-readline --without-zlib CC=aCC

So now I'm trying 8.0.3 with the same arguments. Everthing compiles and installs, initdb seems to run ok, but when I start the postmaster I get a bunch of errors...
LOG: ordering error in dattktbl table: "ty" >= ""
LOG: ordering error in dattktbl table: "" >= ""

So I built with debug and checked it out...
In datetime.c, function CheckDateTokenTables
() the call to CheckDateTokenTable(
"datetktbl", datetktbl, szdatetktbl) seems correct, but once I step into that function, the parameter datetktbl seems to be incorrect. In CheckDateTokenTables
the memory locat
ion of datetktbl is 0x4008ec50, while in CheckDateTokenTable the memory location is 0x400.

So I tried compiling the whole thing with cc instead of aCC and postmaster just cores (I haven't thrown the debugger on yet).

Any clues as to what I should try? (I've been perusing the emails for HP ports, and I haven't found my particular issue - I hate it when that happens!).

Thanks in advance

____________________
____________________
__________
Do You Yahoo!?
Tired of spam? Yahoo! Mail has the best spam protection around
http://mail.yahoo.com

Tom Lane

2005-07-14, 9:23 am

Mark Richardson <markmapo@yahoo.com> writes:
> So now I'm trying 8.0.3 with the same arguments. Everthing compiles and installs, initdb seems to run ok, but when I start the postmaster I get a bunch of errors...
> LOG: ordering error in dattktbl table: "ty" >= ""
> LOG: ordering error in dattktbl table: "" >= ""

> Any clues as to what I should try?

Get a compiler update from HP, or try reducing the optimization level,
or try gcc. The last few cc releases seem to have been a bit buggy :-(

FWIW, here are my notes from the last time I built Postgres on HP's
testdrive systems, which was about six months ago:

On 11.11:
gcc/main ok
gcc/contrib ok
cc/main ok (with +O2 even)
cc/contrib ok
/usr/local/pa64/bin:
gcc64/main Assert in guc.c fails, else ok
gcc64/contrib ok
cc CFLAGS=+DD64 /main ok
cc CFLAGS=+DD64 /contrib ok

On 11.23 (IA64):
gcc not available?
cc/main ok as long as don't use +O with unpatched compiler
cc/contrib ok
cc CFLAGS=+DD64 /main ok
cc +DD64 /contrib ok

Unfortunately I didn't make a note of exactly which compiler versions
were installed on these machines at the time.

regards, tom lane

---------------------------(end of broadcast)---------------------------
TIP 9: In versions below 8.0, the planner will ignore your desire to
choose an index scan if your joining column's datatypes do not
match

Sponsored Links