Re: sequences implementation for DB/pgsql
| From: | Stig S. Bakken | Date: | Wed, 13 Dec 2000 16:53:58 +0000 |
| Subject: | Re: sequences implementation for DB/pgsql | ||
| References: | 1 | Groups: | php.pear |
| Request: | Send a blank email to php-pear+get-683@lists.php.net to get a copy of this message | ||
DB_pgsql and DB_mysql both use db_seq_${name} now. I think that's a
good standard.
- Stig
Peter Bowen wrote:
>
> Should we be setting a default standard for sequences? If we
> specify an append of _seq for all sequences, then don't we break
> compatibility with current databases, and make life very hard for those
> who use other languages to connect to the same database?
>
> Thanks.
> Peter
>
> On Wed, 13 Dec 2000, Stig S. Bakken wrote:
>
> > Alexey Borzov wrote:
> > >
> > > Greetings!
> > >
> > > I've decided to give PEAR::DB a test and immediately noticed
> > > that DB_pgsql lacks sequence support.
> > > So I copied an implementation from DB_mysql and redone it for native
> > > Postgres sequences.
> > > The functions are tested and work as intended:
> > > 1) if a sequence exists, nextId is returned,
> > > 2) if it doesn't exist - it gets created,
> > > 3) if it cannot be created - an error is returned (I tested with the
> > > sequence names that begin with 'pg_' prefix, which is reserved for
> > > system catalogs).
> > > The functions are formatted in accordance with the "latest
> > > standard", PHPDoc comments are added, blah-blah-blah
> > > So you can just commit them into CVS. ;]
> >
> > Hi Alexey,
> >
> > I have sequences implemented in my local checkout, but I rewrote some
> > error handling stuff at the same time that broke a few tests, so it's
> > not committed yet. :-) I didn't append "_seq" to the actual sequence
> > names though, it's probably a good idea, so I'll add that. Thanks.
> >
> > - Stig
> >
> >