Re: Index creation takes for ever - Mailing list pgsql-hackers

From Bruce Momjian
Subject Re: Index creation takes for ever
Date
Msg-id 200308301511.h7UFBBH04513@candle.pha.pa.us
Whole thread Raw
In response to Re: Index creation takes for ever  (Tom Lane <tgl@sss.pgh.pa.us>)
Responses Re: Index creation takes for ever
List pgsql-hackers
Tom Lane wrote:
> ohp@pyrenet.fr writes:
> > This is on unixware 7 (both 7.3.4 and 7.4b)
> 
> > I'm on the FR language (I'll re-initdb whith lang=C to see what happens)
> 
> Okay.  If you find it's still slow in C locale, the next thing to try
> would be forcing use of our own qsort, as we already do for Solaris.
> You'd need to tweak this bit in configure.in:
> 
> # Solaris has a very slow qsort in certain cases, so we replace it.
> case $host_os in solaris*) 
> AC_LIBOBJ(qsort) ;;
> esac
> 
> I'm not sure why it's done that way though.  'Twould be better to let
> the per-platform template files determine it.

The macro has to be in the configure.in file, but we could flag whether
the macro should be used in the template file.  I have avoided adding a
variable just for one platform because it just adds an extra level of
abstraction for little benefit.

--  Bruce Momjian                        |  http://candle.pha.pa.us pgman@candle.pha.pa.us               |  (610)
359-1001+  If your life is a hard drive,     |  13 Roberts Road +  Christ can be your backup.        |  Newtown Square,
Pennsylvania19073
 


pgsql-hackers by date:

Previous
From: Tom Lane
Date:
Subject: Re: SetVariable
Next
From: Andrew Dunstan
Date:
Subject: Re: Linux2.6 overcommit behaviour