Re: reducing the footprint of ScanKeyword (was Re: Large writable variables)

On 1/7/19 7:52 PM, Andres Freund wrote:
> Hi,
>
> On 2019-01-07 19:37:51 -0500, Tom Lane wrote:
>> Andres Freund <andres(at)anarazel(dot)de> writes:
>>> Hm, shouldn't we extract the perfect hash generation into a perl module
>>> or such? It seems that there's plenty other possible uses for it.
>> Such as?
> Builtin functions for one, which we'd swatted down last time round due
> to gperfs defficiencies. But I think there's plenty more potential,
> e.g. it'd make sense from a performance POV to use a perfect hash
> function for locks on builtin objects (the hashtable for lookups therein
> shows up prominently in a fair number of profiles, and they are a large
> percentage of the acquistions). I'm certain there's plenty more, I've
> not though too much about it.
>

Yeah, this is pretty neat,

>> But in any case, that sounds like a task for someone with
>> more sense of Perl style than I have.
> John, any chance you could help out with that... :)
>

If he doesn't I will.

cheers

andrew

--
Andrew Dunstan https://www.2ndQuadrant.com
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services