I will watch it asap. Anyway, I am not trying to discourage you or oppose to it. If you have resources and want to start building something like this, you have all my support. Some people may love it. What I cannot promise is that it will be included in web2py until I see a real benefit.
On Tuesday, 30 April 2013 15:49:53 UTC-5, Arnon Marcus wrote: > > Hi Massimo, > You should "really" watch this lecture: > http://lanyrd.com/2011/pygotham/shpkm/#link-fkfb > > I think you will like it, even if you don't like ORMs. > It shows some architectural diagrams that you might find interesting, for > both the Core layer and ORM layer of SQLAlchemy. > > I think this would make clear the dichotomy between the two. > > As the lecture shows, the Core and the ORM are actually separate > code-bases (though the ORM is built on-top of Core), so that people can > actually not use the ORM at all if they don't like to, and what they will > be left eith, is basically equivalent to what web2py provides - a DAL. > > Since these are separate layers, I think what might be an interesting > research-project, is not re-inventing the wheel, but instead try to layer > SQLA's ORM on-top of web2py's DAL. The past-experiments I've seen where > trying to do the opposite, but I actually think that this makes more sense. > > The modern way of generating an ORM class, is called the > "declarative-mapping" approach, in which you build the schema right into > the python classes ("decleratively"). This would be way to difficult to > re-instrument. > > However, there still exist the 'older', more "dis-jointed" way of mapping > that they call "classical-mapping", which maps a completely regular > python classes to a 'Core' schema-object, and "generates" an ORM-class out > of the two using a 3'rd object called a "mapper". > > It is basically taking a regular python class, and a schema that was built > using Core (in an almost identical fashion as it is done in the web2py-DAL) > and using a separate mapper-object to 'instrument' the schema into the > class, and thus generating an ORM class out of the merge of the two. > > So I was thinking it might be interesting research-project, is to just try > to sub-class the mapper-object, and write a web2py-DAL-mapper variation, > that basically generates the same ORM, using the same kind of > instrumentation, but just using the web2py-DAL schema-object instead of the > SQLA-Core one. > > What do you say? > Is that feasible? > -- --- You received this message because you are subscribed to the Google Groups "web2py-users" group. To unsubscribe from this group and stop receiving emails from it, send an email to web2py+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/groups/opt_out.
