Re: [Phpwebsite-developers] Database abstraction layer

[Karsten D. <k.d...@tu...>]

On Sun, Mar 04, 2001 at 03:10:08PM -0600, Richard Rowell wrote:
> On Saturday 03 March 2001 17:11, you wrote:
> > Richard, I tend to agree with Karsten that we should use PEAR or someth=
ing
> > else "standard", if there is such a thing, or phplib.
>=20
> My major conecerns with PEAR are:
> A.  It is mosty vapor at the moment. (check for yourslef http://pear.php.=
net/)

Well, as far as I got it, the code is more advanced than documentation and
website.

> B.  The write up I saw on the PHPBuilder site
> (http://www.phpbuilder.com/columns/allan20010115.php3)

Quickly read that, and I think it sounds fairly good... But I did not read
the comments on the article, maybe there is some hard criticism in them.
More worries are caused on my side by an email on the phplib mailing list I
got today (phplib will be merged into PEAR, that is the plan so far). Let me
quote a little bit:

=3D=3D=3D=3D=3D=3D=3Dsnip=3D=3D=3D=3D=3D=3D
> Yes, there are some problems with PEAR (and PHPLib) now. PEAR seems to
> choose not the right way (IMHO) - making well-organized (that's well),
> but almost unusable stuff. PEAR::DB (the main thing in the PEAR now,
> that could be used in the real life) is obviously too slow compared to
> the PHPLIB counterpart, missing obvious things (like affected_rows,
> which almost every DB supports, num_rows for Mysql, for example).  =20
> I consider it a bloatware (sorry, Stig).

I buy nearly all of this, the PEAR stile tends to overdo things and make
code too slow. I currently have problems to keep out all that from the
new cache stuff as people already ask for more hidden addslashes() when
saving cache data to and for error handling 1 + 1... But instead of
complaining, who will tell this to Stig and Chuck? Who will start the
argue?
=3D=3D=3D=3D=3D=3D=3D=3Dsnip=3D=3D=3D=3D=3D=3D

The unsupported stuff is probably going to be added and/or fixed over time,
but the speed issue is something that worries me. More comments on that see
below...

> C.  Almost zero documentation (relates back to A)

That's a point.

> I haven't looked at PHPLib, I will do so and venture an opinion.

OK, as far as phplib is concerned: it works very well, is quite complete,
simple and fast. But it is scheduled to be merged into PEAR (as mentioned
above), so it might not be the best idea to use it for a (new) project.

Has anyone looked at metabase by manuel Lemos? It can be found at
phpclasses.upperdesign.com and claims to be more or less the db abstraction
layer with the most complete feature set and has some very nifty features
(like XML based database description, allowing to initialize all supported
databases from on XML description file. This makes all those xxxxx.sql file
superflous). But I didn't really look into this yet.

Bye,
Karsten
--=20
Why do we have to hide from the police, daddy?
Because we use emacs, son. They use vi.
-----------------------------
mailto:k.d...@tu... w=B3: http://www.k-fish.de/
gpg: http://www.k-fish.de/mykeys.gpg