Обсуждение: Re: Antwort: [HACKERS] ecpg parser

> Right now preproc.y and gram.y are in
> sync in the cvs tree. If you change anything please send me the diff
> or tell me it has changed so I don't miss it. I have an old gram.y
> version around so I can diff it locally.

I haven't yet committed the changes to the source tree, but may have a
chance to regression test and commit today. I'll send patches to you at
the same time. btw, there was a missing comma between ADD and AS in the
token list (I see the problem also in preproc.y) and I found one or two
omissions in parser/keywords.c. Most of my changes were to fix omissions
and to allow tokens to be column identifiers or column labels as much as
possible.

The other thing I'm doing is documenting the reserved and unreserved
keywords (I'm most of the way there, and will commit a new file
syntax.sgml to the docs tree soon). The only thing I've been looking at
so far is gram.y. Are there any other keywords added to preproc.y? I see
that you did things so that EXEC and SQL are not required to be reserved
words (I hadn't thought of how to do that; nice trick!), but perhaps
there are others which are. Shall we put all of the keywords used by
either gram.y or preproc.y into gram.y and keywords.c to make it more
consistant and to enforce consistancy in database design? SQL92 doesn't
seem to make a big distinction between embedded and interactive SQL wrt
keyword restrictions...

> Please send mail to me at Michael.Meskes@usa.net. That
> way it's guaranteed to reach m.

OK, sorry. I'd put your address into my mailer, and hadn't notice the
change.

                       - Tom

On Tue, Sep 29, 1998 at 12:55:57PM +0000, Thomas G. Lockhart wrote:
> I haven't yet committed the changes to the source tree, but may have a
> chance to regression test and commit today. I'll send patches to you at
> the same time. btw, there was a missing comma between ADD and AS in the
> token list (I see the problem also in preproc.y) and I found one or two
> omissions in parser/keywords.c. Most of my changes were to fix omissions
> and to allow tokens to be column identifiers or column labels as much as
> possible.

I see. I will add your changes as soon as I find time.

> The other thing I'm doing is documenting the reserved and unreserved
> keywords (I'm most of the way there, and will commit a new file
> syntax.sgml to the docs tree soon). The only thing I've been looking at
> so far is gram.y. Are there any other keywords added to preproc.y? I see

Yes, there are some special embedded sql keywords like break, call,
whenever, etc. See ecpg/preproc/ecpg_keywords.c for details.

> that you did things so that EXEC and SQL are not required to be reserved
> words (I hadn't thought of how to do that; nice trick!), but perhaps

Thanks. :-)

> there are others which are. Shall we put all of the keywords used by
> either gram.y or preproc.y into gram.y and keywords.c to make it more
> consistant and to enforce consistancy in database design? SQL92 doesn't

At first I thought why making it a keyword when it doesn't have to. But the
more I think about it the more I like it. It just doesn't make sense to use
these keywords because that would force you to not use ecpg.

> OK, sorry. I'd put your address into my mailer, and hadn't notice the
> change.

No problem. But that way it will reach me no matter if I'm at home or in the
office.

Michael
--
Dr. Michael Meskes      | Th.-Heuss-Str. 61, D-41812 Erkelenz | Go SF49ers!
Senior-Consultant       | business: Michael.Meskes@mummert.de | Go Rhein Fire!
Mummert+Partner         | private: Michael.Meskes@usa.net     | Use Debian
Unternehmensberatung AG |          Michael.Meskes@gmx.net     | GNU/Linux!