Related articles |
---|
Parsing SQL'92 terryvb@my-deja.com (2000-05-28) |
Re: Parsing SQL'92 derekNOSPAM@knosof.co.uk (2000-05-30) |
From: | derekNOSPAM@knosof.co.uk (Derek M Jones) |
Newsgroups: | comp.compilers,comp.databases |
Date: | 30 May 2000 02:35:30 -0400 |
Organization: | Knowledge Software |
References: | 00-05-098 |
Keywords: | SQL, parse |
All,
>I've been trying to parse the SQL'92 grammar using various parser
>generators, but so far no luck. I've thrown yacc (LALR(1)), yay (LALR
>(2)), and JavaCC (LL(k)) at it, but I always get a ton of conflicts.
After a months work I managed to get it down to 2 conflicts.
>Can someone answer definitively what language category SQL'92 is in,
Well it can be made LALR(1), if you are willing to give the generate some
hints on how to handle the conflicts.
>The grammars I've used were generated by processing the BNF from the
>ISO standard document.
That is where I started.
After SQL/2 you have the vendor extensions to worry about. If you
want to support more than one vendor then the conflicts really start
to mount up (and there are significant lexical differences).
On a commercial note (being a commercial outfit) people make a lot of
noise about SQL/2 grammars, but don't spend any money (either doing it
themselves or buying mine) doing anything about it. Do people look
at SQL for fun, discover it is hard and decide to go and do something
easier?
Return to the
comp.compilers page.
Search the
comp.compilers archives again.