interix port running...

From: "Merlin Moncure" <merlin(dot)moncure(at)rcsonline(dot)com>
To: "pgsql-hackers-win32" <pgsql-hackers-win32(at)postgresql(dot)org>
Subject: interix port running...
Date: 2004-02-03 20:32:43
Message-ID: 303E00EBDD07B943924382E153890E5434AA78@cuthbert.rcsinc.local
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers-win32

Merlin.Moncure=# select version();
version
---------------------------------------------------------------------
PostgreSQL 7.4.1 on i386-pc-interix3, compiled by GCC gcc (GCC) 3.3
(1 row)

Merlin.Moncure=#

I never could get initdb to work, so I had to borrow initdb from the
cygwin compiled version. Also, there were a few quirky problems with
the compile/build I had to work around. Since I'm new to Unix, this was
a fun learning exercise. I've been administrating postgresql servers
for several years now, but have never bothered to learn unix much beyond
running initdb :).

First problem: no config.guess
I downloaded postgresql 7.2.2 port to interix and copied all the interix
specific stuff...this fixed it.

Second problem: configure fails to write pg_config.h. Actually,
configure fails to write any *config.h on my machine for some reason.
This appears to be somewhat uncommon, but it was a huge PITA (compiling
anything is a battle). I manually copied all defines form config.log to
pg_config.h. I wasted a lot of time initially until I realized I could
do this.

Third problem: no sync()
Hacked around it with #define sync() in interix3.h...ignoring this one
for now.

Fourth problem:
Make install fails on...can't find ascii_and_mic.so. Couldn't figure
this one out so just did
gmake -C src/bin install
this went through ok.

Fifth problem:
Command line tools won't work...just used cygwins...mingw compiled tools
probably will work ok to.

Final thoughts...some (very) rough comparisons with the cygwin version
via pgbench show that the interix version is about 10% slower. This is
really interesting and a little disappointing. However, everything
appears to function ok (haven't run regression yet). I think the
current efforts on the win32 port are still quite justified because of
lower installation overhead. Also, the freedom to execute native API
calls leaves a lot of room for optimization as the win32 API is quite
sophisticated, if somewhat esoteric, in this regard.

Can anybody suggest a good command line debugger/profiler?

Browse pgsql-hackers-win32 by date

  From Date Subject
Next Message Magnus Hagander 2004-02-03 23:46:02 New win32 signals patch (3)
Previous Message Tom Lane 2004-02-03 15:05:34 Re: [pgsql-hackers-win32] Sync vs. fsync during checkpoint