Thread: Two question about performance tuning issue.
Hi, I'm a newbie to postgresql, and I have 2 questions.
1. shared memory hash key is string.
why don't we use predefined integer value?
I think it is dangerous to use string for key value,
because it can be mistyped , which might lead us to debug more and more.
why don't we use predefined integer value?
I think it is dangerous to use string for key value,
because it can be mistyped , which might lead us to debug more and more.
2. appendStringInfoChar reallocates every bytes whenever needed,
why don't reallocate chunk for better performace?
because initStringInfo allocates only 256 bytes ,
pg_beginmessage prepared 256 bytes of StringInfo for communication buffer.
why don't reallocate chunk for better performace?
because initStringInfo allocates only 256 bytes ,
pg_beginmessage prepared 256 bytes of StringInfo for communication buffer.
It's hard to analyze pgsql source code,
because I'm a newbie, and have no idea about postgres internals,
But I will try hard!!
Thank you.
Re: Two question about performance tuning issue.
From
nconway@klamath.dyndns.org (Neil Conway)
Date:
On Fri, Jul 26, 2002 at 03:43:30AM +0900, Kangmo, Kim wrote: > 2. appendStringInfoChar reallocates every bytes whenever needed, > why don't reallocate chunk for better performace? enlargeStringInfo() takes care of that -- from the comment in the function: /* * We don't want to allocate just a little more space with each * append; for efficiency, double the buffer sizeeach time it * overflows. Actually, we might need to more than double it if * 'needed' is big... */ Cheers, Neil -- Neil Conway <neilconway@rogers.com> PGP Key ID: DB3C29FC