All the Perl that's Practical to Extract and Report

Modules I like

acme on 2002-07-19T08:58:26

The slides for my Modules I like, Modules I hate talk I presented at yesterday's London.pm technical meeting are available. More details soon.

many kinds of POOP

autarch on 2002-07-19T16:21:27

I just sent this as an email but then I realized that I'd rather post it here for all the world to see ;)

Lumping Alazbo in with SPOPS and Tangram is highly misleading. Methinks you haven't actually looked at what it does.

You say '"Almost, but not quite" e.g. schemas change'

Duh!

Alzabo is a _data modelling_ tool as well. If your schema needs to change you can use the data modelling tool to change it. Simple, easy, quick.
And then you can ask it to generate the SQL needed to update your RDBMS with the new schema (ALTER TABLE ...). Or you could change your schema in
the RDBMS and tell Alzabo to sync itself with the RDBMS. Also quite simple.

And once your schema changes, it can _still_ auto-generate (on the fly, not to a file) methods that are appropriate for your schema.

As to needing a "proper object database" for Perl, IMHO, an object database is useless for many reasons, but if you did want one, that's what Tangram is!

Re:many kinds of POOP

acme on 2002-07-19T16:28:50

[oh, then i guess i'll post my reply here too]

Oh, sorry, the slides didn't make it quite clear. In the actual talk I did say that Alzabo handles schema changes and generally gets closer to what we want it to do.

I'm afraid Alzabo still doesn't hit my personal sweet spot, so it stays there ;-)

[about tangram being an oodb] No it's not. It just pretends to be. Something like Pixie, say is a proper object database. Unfortunately, object databases tend to be slow. Oh well.

Re:many kinds of POOP

autarch on 2002-07-19T17:03:52

But if you need to do actual queries, which _many_ applications need to do, you can't really use Pixie.

If you only ever need to look things up by OID, and you _already_ know the OID of the object(s) you want, Pixie fits the bill. But that's a _very_ narrow way to get at data.

This hardly sounds like a "database" to me, since it has no facilities for querying whatsoever.

A real OODBMS (for some values of "real") would simply allow you to call object/class methods and would transparently persist objects. As to why _this_ too is a bad idea, this article goes into more detail than I can here.