Yeah, something like that I would think. Is there a generic set of queries that some/most/all of certain types of NoSQL DBs share? Mongo is nice to support but also Postgres, Redis, etc.
Are there some queries where DAL-style syntax makes sense? For example selecting for an item or set of items (without a join, I suppose)? And then being able to turn any result sets into Rows object would be good. On Monday, November 26, 2012 6:23:10 AM UTC-8, rochacbruno wrote: > > I think DAL has nothing to do with Mongo. > > It is much easier to use only PyMongo API. > > What we really need is a "Rows factory" it shoul take arbitrary data > format (can have a scheme) and will give us back a Rows object. Also it > should have some event binders to update, delete, insert... > > So programmers will do this: > > mydata = pymongo.whatever() # pymongo dict like objects > > rows = gluon.RowsFactory(mydata, scheme=myscheme, updater=lambda, > inserter=lambda, deleter=lambda) > > now we should have a "Rows" object and it could be used for grid, forms > etc... > > The problem is that it can take long time to convert things in to Rows. > > So the best should be a dict + events based Form, grid and other controls > designed specifically for these cases. > > Bruno Rocha > http://rochacbruno.com.br > mobile > Em 26/11/2012 04:06, "Mark Kirkwood" <mark.k...@gmail.com <javascript:>> > escreveu: > >> I am wondering if there are some deeper issues getting NoSQL support in >> Web2py - for instance the model part seems pretty tied into relational >> ideas (e.g db.define_table and ..references db.dog etc). This design works >> really well and insulates developers from much of the fiddlyness of dealing >> directly with relation db's quirks, but the ideas themselves are clearly >> relational. >> >> The other big DAL issue would be joins..but I guess integration with GAE >> has covered that to some extent (tho it looks like GAE can kinda fake a >> join with referenced properties...might be able to do something like that >> with Mongo, but not for all other NoSQLs). >> >> On Wednesday, November 21, 2012 11:58:35 PM UTC+13, Niphlod wrote: >>> >>> that is exactly the explanation of the term "experimental". >>> The problem as always is that if noone starts to test it, it will be in >>> this way forever. >>> >>> On Tuesday, November 20, 2012 11:25:59 PM UTC+1, Simon Ashley wrote: >>>> >>>> The bottom line seems to be that we/ others need to start to use nosql >>>> engines to sort the issues. >>>> My guess is that, currently, critical mass is not present to rely on it >>>> for production sites. >>>> >>>> -- >> >> >> >> > --
