[development] Oracle porting & db_num_rows patches

[Gabriele Turchi]

Il giorno lun, 09/04/2007 alle 12.31 -0500, Larry Garfield ha scritto:
> On Sat, 7 Apr 2007 23:26:55 +0200, Dries Buytaert <dries.buytaert at gmail.com> wrote:
> > Hi Gabrieli,
> > 
> > On 07 Apr 2007, at 10:40, Gabriele Turchi wrote:
> >> Hi all. I'm doing a drupal 5.1 Oracle db experimental port. I have a
> >> first core installation more or less working.
> >>
> >> Next target will be cck and views.
> >>
> >> One problem was the need to remove the db_num_rows calls
> >> (unavailable in
> >> Oracle). I've seen here and there an inappropriate use of this
> >> function
> >> ("select *" and db_num_rows in place of "select count(*)"), and I
> >> think
> >> my patches can be useful also for performance under other databases.
> > 
> > Cool! :)
> > 
> > Note that we can't commit your patch to the Drupal 5 branch -- only
> > bugfixes are committed to that branch.  You'll want to provide
> > patches for CVS HEAD so that the new functionality can become part of
> > Drupal 6.
> > 
> > You'll want to check with Larry Garfield -- he's working on PDO
> > support for Drupal 6, and is also looking into removing db_num_rows
> > ().  See this week's thread about "PDO and database limitations" for
> > more information.
> 
> @Gabrieli: Greetings!  Hopefully what I'm working on will make Oracle 
> support easier, too.

I think that too.

> 
> @Dries: Since it looks like db_num_rows() is going to be an issue for 
> more than just PDO, would you be open to separate patches that just 
> do "SQL cleanup" to eliminate db_num_rows(), clean up 
> tablename-in-prepared-statement problems, etc, either as lots of little 
> patches (my preference, easier to test/debug) or a big cleanup patch?  
> I believe that will be the largest remaining part of PDO anyway, but 
> valuable to native-Oracle efforts etc. as well.
> 

To be honest, the big part of my patches about db_num_rows are
one-liner... (change the select using "count(*)" where only the record
number is needed and turning upside-down a while loop when fetching
data).

Maybe, creating some sort of "semi-official" patches should be useful to
do something like a "beta-test" during the "5." release.

About changes, may I also add a "we don't really need a kilometer-long
table and index names"? Oracle doesn't like that...

> --Larry Garfield
> 

Best regards,
Gabriele

P.S.: I'm sorry, my english is alpha version...
-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/x-pkcs7-signature
Size: 6054 bytes
Desc: not available
Url : http://lists.drupal.org/pipermail/development/attachments/20070410/32932153/attachment-0001.bin