2009/8/14 <csrabak@bol.com.br>
Esteban,
I'm comenting this on *philosophical* grounds.
I understand the efforts to have SqueakDBX and other ORM in Pharo are motivated by Seaside and other "production" initiatives so, again, my food for thought is more an intellectual contribution which I see as useful for Pharo as project:
I don't see Pharo *any time soon*® having all the toolset to be able to compete in the CRUD¹ realm with more streamlined tools and besides, not matter how much theoretical work has been done on ORM, the "impedance mismatch" is still there,
Hi! I am agree with you about the "impedance mismatch". I think that if I am free to choose a persistence strategy I would use an OODB. However, if we are talking about real enterprise application (not a simple webpage) being able to choose the persistence strategy is practically impossible. We did a survey last year and the results were that most of the times you cannot choose. Of course, the client has a lot of reasons: - The client already has a RDBMS - They had paid for it and have the license (sometimes) - They want a company's support. The only company here is Gemstone. - RDBMS has history and it is an standard - They have knowledge about it - Interaction with other system (even legazy systems) - They are afraid of using another persistence strategy - They want to use SQL - They have DBAs - The persistence is difficult to negotiate So. In my opinion, if Pharo wants to be used really as a platform for enterprise applications (competing to java, .NET, etc), it must have a good relational solution. Of course, when you are able to choose, you can go for another approach like OODB. Best, Mariano
a recent and practicall account of this (for Squeak) can be found in
http://onsmalltalk.com/simple-image-based-persistence-in-squeak
also, AIDA/Web has been able to run only storing thing in the image! See
http://groups.google.com/group/comp.lang.smalltalk/browse_thread/thread/d3ba...
Since we'll increase the educational efforts to spread Pharo, I think that as technology we should to recover a bit of Smalltalk technology and be first more Object Oriented and then see if ORM still is so needed.
my 0.0199999....
--
Cesar Rabak
[1] http://en.wikipedia.org/wiki/Create,_read,_update_and_delete Em 14/08/2009 16:14, *Esteban A. Maringolo < emaringolo@gmail.com >*escreveu:
My prototype is getting less proto, and I found out it doesn't have complex relations, and the system is going to perform faster if I have tables for most of it.
What are the choices I have for doing ORM in Pharo?
¿Does GLORP work? ¿Any other options? I don't have a strong preference for the RDBMS engine, it can be anything (free), being it MySQL, PostgreSQL or Sql Server Express.
Best regards,
Esteban A. Maringolo
_______________________________________________ Pharo-project mailing list Pharo-project@lists.gforge.inria.fr http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project
_______________________________________________ Pharo-project mailing list Pharo-project@lists.gforge.inria.fr http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/pharo-project