domingo, abril 09, 2006

Más sobre MDA vs SF

Continuando una discusión abierta hace algún tiempo en The Server Side, Jack Vaughan expone sobre Software Factories brevemente, sólo para recordar que parecería que el principal objetivo de hablar sobre SF es desacreditar la iniciativa Model Driven Architecture. Sin embargo, en el marco de usuarios esperanzados en Visual Studio 2005 Team System, un puñado de observaciones críticas apuntan a la inconsistencia de este concepto en su desarrollo actual: en primer lugar, simplemente el remate del comentario de Vaughan.
Not discussed here, of course, is the idea that a key goal behind MDA is to create models that are high-level enough to be platform independent. Frankly, and for obvious reasons, this is not too often a driver in the MS camp
...es decir, evidentemente un punto fundamental en la discusión es la posibilidad de modelar con independencia de la plataforma, bajando el diseño a código, sea en Windows, Linux, Unix o AS400.
Pero es en la pequeña lista de respuestas, casi todas ellas desde el propio campo, donde se puede sacar más sustancia:
Dice Tad Anderson:
My frustration is with today, and the tools we are being handed today. With VSTS 2005 I lose XDE, I gain a DSL roundtrip class diagram modeling tool, and a bunch of Data Center modeling tools. So now I am stuck with the choice between Sparx, Borland, and some other tools to be able to do my job today. I don't have time to write a DSL UML tool so I can architect this project, and I wouldn't even if I had the time.
My frustration is not with your Software Factory movement, my frustration is with the gap that has been put in place by MS making us wait for your Software Factories movement to produce the tools and process frameworks we need to do our job. So I am not against your movement of Software Factories, I am against Microsoft's position of promoting it as what will be available, while providing nothing to us today
Otras intervenciones incluyen una defensa de SF de Aidas Ozelis, respondida detalladamente por Jon Kern, de OptimalJ.

No hay comentarios.: