db abstraction practice

[miasma]

Looking at some stuff, it seems that a good thing to start practicing is doing some database abstraction...

I've been looking at some examples (PHPLib's db_mysql , some random code here and there)..

One thing that I'm a little unsure of is the following..

say I had a situation where I wanted to grab two different result sets from the database... (say I wanted to grab a bunch of news records from one table... and I wanted to grab say a color scheme from another table)...

It looks like I have two options:

a) I would need to create two instances of my DB class.. which would create two separate connections. This doesn't seem to be good practice to me... am i wrong?

b) I would need to create one instance, perform a query.. store the records in an array.. and then perform a second query.. store the records in an array... and then later on in the page cycle through the array.

Which is the more performed practice? And is there another way that I'm not thinking of?

[kparker]

You should be able to have multiple queries active on the same db connection at the same time. Are there some dbms that don't support this behavior?

[miasma]

well.. I'm talking about using db abstraction in a class.

For instance.. in mysql.. surely you can pass the connection identifier to a mysql_query and get a resultset back... but every dbabstraction class I see only has one connection identifier and one result identifier... in other words, doing only one query/result at a time.

[kparker]

Really? They all keep the result ID as a hidden internal member? Yikes, no wonder I never found one I liked...

[miasma]

What do you do instead for database abstraction? a purely functional version? just a few functions?