Re: [HACKERS] gperf anyone?

* Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> [000118 21:10] wrote:
> > At 07:36 PM 1/18/00 -0500, Bruce Momjian wrote:
> > I wondered about this last, i.e. the use of GNU code since Postgres
> > is licensed differently.
>
> AFAIK this is no worse than using flex or bison --- the source code of
> gperf is GPL'ed, but its output is not.
>
> Don Baccus <dhogaza(at)pacifier(dot)com> writes:
> > Whether faster or slower, though, I can't imagine either method taking
> > noticably more than 0% of the total time to process a query, even the
> > most simple queries.
>
> I agree with Don that the performance benefit is likely to be
> unmeasurable. Still, there could be a win: we currently have to modify
> keywords.c by hand every time we have to add/delete a keyword. Does
> gperf offer any aid for maintaining the keyword list? If so, that'd
> be sufficient reason to switch to it...

any minimal performance boost shows over time, unfortunatly using gperf
will require that you either:

a) require gperf to be installed on a system that compiles postgresql
b) manually maintain the gperf compiled files in your CVS repo
(sort of like syscalls in FreeBSD)

Option B is not that bad at the expense of additional contributor
overhead.

I hope to be able to present some soft of bench to determine
if gperf is worth the additional effort of maintainance in the
near future.

in the meanwhile, happy hacking. :)

--
-Alfred Perlstein - [bright(at)wintelcom(dot)net|alfred(at)freebsd(dot)org]