Re: memory: bug or feature - Mailing list pgsql-hackers

From Tom Lane
Subject Re: memory: bug or feature
Date
Msg-id 783.963242928@sss.pgh.pa.us
Whole thread Raw
In response to memory: bug or feature  (Karel Zak <zakkr@zf.jcu.cz>)
List pgsql-hackers
Karel Zak <zakkr@zf.jcu.cz> writes:
>  I'm playing with MemoryContextCheck() and I a little confuse now.
>  How idea is for:
>     context->initBlockSize
>    context->maxBlockSize

> Must be a memory block already between this range?

No, those are just hints for allocation of default-sized blocks.  They
don't constrain allocation of blocks that have to be a particular size
to accommodate a large chunk.  The reason for having them is just to
allow a caller who expects that a particular context won't contain much
data to prevent a lot of wasted space from being allocated for big
default-sized blocks.
        regards, tom lane


pgsql-hackers by date:

Previous
From: JanWieck@t-online.de (Jan Wieck)
Date:
Subject: Re: Re: [SQL] Re: [GENERAL] lztext and compression ratios...
Next
From: Ed Loehr
Date:
Subject: Re: Re: postgres TODO