Re: ECPG won't compile anymore

> -----Original Message-----
> From: Dann Corbit
> Sent: Tuesday, June 18, 2002 1:51 PM
> To: Bruce Momjian; Michael Meskes
> Cc: PostgreSQL Hacker
> Subject: Re: [HACKERS] ECPG won't compile anymore
>
>
> > -----Original Message-----
> > From: Bruce Momjian [mailto:pgman(at)candle(dot)pha(dot)pa(dot)us]
> > Sent: Tuesday, June 18, 2002 11:13 AM
> > To: Michael Meskes
> > Cc: PostgreSQL Hacker
> > Subject: Re: [HACKERS] ECPG won't compile anymore
> >
> >
> > Michael Meskes wrote:
> > > On Tue, Jun 18, 2002 at 10:29:10AM -0400, Tom Lane wrote:
> > > > I'd be inclined to say that you don't commit until bison 1.49 is
> > > > officially released. Got any idea when that will be?
> > >
> > > No, that's the problem. ECPG and the backend parser are
> > running out of
> > > sync. After all bison's release may be later than our next one.
> > >
> > > I cannot commit even simple bugfixes anymore as my source tree
> > > already has the uncompilable bison file. So I would have to
> > work on two
> > > different source trees. I don't exactly like that.
> >
> > Are we the only ones up against this problem? Hard to
> imagine we are
> > the only ones up against this limit in bison. Are there
> > other options?
> > I don't see how we can distribute ecpg in 7.3 without some
> > kind of fix.
>
> There are some other freely available parser/generators. I like the
> Lemmon parser generator. Of course, I have no idea how traumatic it
> would be to convert a Bison grammar into Lemmon. There is also PCCTS
> and some other free ones.
>
> http://www.hwaci.com/sw/lemon/
>

It occurs to me that SQLite is a PostgreSQL clone {grammar wise, but a
subset}:
http://www.hwaci.com/sw/sqlite/
that uses the Lemon parser generator. Therefore, the grammar for the
SQL language itself should be extremely similar, and it might
(therefore) be very easy to see what he has done to make the transition.
Of course, the ECPG tool has its own grammar, so I am not sure how
helpful that would be.

By the way, there is a rather unflattering speed comparison with
PostgreSQL on this page:
http://www.hwaci.com/sw/sqlite/speed.html

It might be nice to use those tests with gprof to find out where the
bottlenecks are. It also seems possible that he may have used an older
version of PostgreSQL.