Re: XML_Indexing and DB_Dataobject
| From: | Justin Patrin | Date: | Tue, 05 Oct 2004 23:59:07 +0000 |
| Subject: | Re: XML_Indexing and DB_Dataobject | ||
| References: | 1 2 3 | Groups: | php.pear.dev |
| Request: | Send a blank email to pear-dev+get-33666@lists.php.net to get a copy of this message | ||
On Wed, 06 Oct 2004 00:46:53 +0200, Olivier Guilyardi <ml@xung.org> wrote:
> Justin Patrin wrote:
>
> >>What do you think about XML native databases management with PHP ?
> >>Do you believe XML_Indexing integration with DB_Dataobject is a good idea ?
> >
> > Well, you could create a Data Driver for DB_DataObject, but that seems
> > not too useful to me. It would create more complexity without adding
> > too much usefulness. Can you think of many other data backends?
>
> Are you talking about a whole new "Data Driver" layer ? No, I don't
> think this is a good idea, but anyway that is Alan Knowles to decide.
> I don't know the DB_Dataobject code enough to say how to implement
> XML containers.
>
You say you don't want a data driver, but you still want to use
DataObject with XML? Ir do you mean making an XML_DataObject?
> Apart from XML I don't see many other data backend, no.
>
> > IMHO if you want to store data in XML, it's much more useful to make a
> > DB driver which stores its data in XML. It can even just support a
> > subset of SQL if you want. This way, it takes no change to
> > DB_DataObject and other DB-using projects could make use of it with
> > minimal changes.
> >
> > Of course, this would require SQL parsing....but if you only do a
> > subset, you should be fine. There's also SQL_Parser:
> > http://pear.php.net/package/SQL_Parser.
>
> Well, SQL parsing is heavy... I recognize a DB XML driver is a nice idea.
> If it's just a small subset, then it may be acceptable, yes. Thanks for
> this idea :-)
>
> But, what I like about Dataobjects is that they sufficiently abstract
> the database, so that you interact with it using methods (find(), etc...).
> No need to parse SQL, no overhead. With a such interface, in theory you
> can use whatever data container you wish.
Sure, but DB_DataObject is currently highly tied to DB, hence its
name. It's also highly tied to SQL. A new DataObject could be made
with a data driver backend to support either way, but, as we both have
said, this would be overkill. If you want to make an XML_DataObject, I
don't think anyone would mind.
>
> > This also may make XML_Indexer
> > less useful, but it seems like the right place to stick this
> > functionality to me.
>
> It does not change anything about XML_Indexing usefulness : wether you
> parse SQL or not, you still have to retrieve and update the data
> efficiently (!= loading the whole XML file into memory).
>
--
paperCrane --Justin Patrin--