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

Hi,

On 2019-01-07 16:11:04 -0500, Tom Lane wrote:
> I wrote:
> > I took a quick look through the NetBSD nbperf sources at
> > http://cvsweb.netbsd.org/bsdweb.cgi/src/usr.bin/nbperf/
> > and I concur with your judgment that we could manage translating
> > that into Perl, especially if we only implement the parts we need.
>
> Here's an implementation of that, using the hash functions you showed
> upthread. The speed of the Perl script seems to be pretty acceptable;
> less than 100ms to handle the main SQL keyword list, on my machine.
> Yeah, the C version might be less than 1ms, but I don't think that
> we need to put up with non-Perl build tooling for that.
>
> Using the same test case as before (parsing information_schema.sql),
> I get runtimes around 3560 ms, a shade better than my jury-rigged
> prototype.
>
> Probably there's a lot to be criticized about the Perl style below;
> anybody feel a need to rewrite it?

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.

Greetings,

Andres Freund