On Wed, Jan 09, 2019 at 02:04:15PM -0500, Tom Lane wrote:
> Also, in view of finding that the original multiplier choices failed
> on the fmgr oid problem, I spent a little effort making the code
> able to try more combinations of hash multipliers and seeds. It'd
> be nice to have some theory rath
On Tue, Jan 08, 2019 at 05:53:25PM -0500, Tom Lane wrote:
> John Naylor writes:
> > -As for the graph algorithm, I'd have to play with it to understand
> > how it works.
>
> I improved the comment about how come the hash table entry assignment
> works. One thing I'm not clear about myself is
>
On Sun, Jan 06, 2019 at 02:29:05PM -0500, Tom Lane wrote:
> * It's too bad that the hash function doesn't have a return convention
> that allows distinguishing "couldn't possibly match any keyword" from
> "might match keyword 0". I imagine a lot of the zero entries in its
> hashtable could be inte
On Mon, Jan 07, 2019 at 03:11:55AM +1300, David Rowley wrote:
> What I'm most interested in is how long it took to generate the hash
> function in hash2.c?
It's within the noise floor of time(1) on my laptop, e.g. ~1ms.
Joerg
On Fri, Jan 04, 2019 at 02:36:15PM -0800, Andres Freund wrote:
> Hi,
>
> On 2019-01-04 16:43:39 -0500, Tom Lane wrote:
> > Joerg Sonnenberger writes:
> > >> * What's the generator written in? (if the answer's not "Perl", wedging
>
On Fri, Jan 04, 2019 at 03:31:11PM -0500, Tom Lane wrote:
> John Naylor writes:
> > On 1/3/19, Joerg Sonnenberger wrote:
> >> I was pointed at your patch on IRC and decided to look into adding my
> >> own pieces. What I can provide you is a fast perfect hash fun
On Sun, Dec 16, 2018 at 11:50:15AM -0500, John Naylor wrote:
> A few months ago I was looking into faster search algorithms for
> ScanKeywordLookup(), so this is interesting to me. While an optimal
> full replacement would be a lot of work, the above ideas are much less
> invasive and would still h
