Sessions can use its db driver like so: del db.sessions[0] -> clear table del db.sessions[id] -> drop row db.sessions[0] = db.sessions(**attrs) -> insert db.sessions[id] -> fetch db.sessions[id] = attrs -> update (could contain lambdas for txn)
For example, if you wanted to make an FSDB db driver to store sessions on disk: simply implement: __call__,__getitem__,__setitem__,__delitem__ Robin On Jan 23, 9:14 am, Robin B <[email protected]> wrote: > Now that db.table supports getitem,setitem,delitem, session can work > by only using the new functions, then session db drivers do not need a > query interface, and can be super simple. > > another shortcut: > > del db.posts[0] -> drop table > > I still think it is useful to get the default values: > db.posts(**attrs) > > posts = db.posts > posts[0]=posts(**attrs) > > Robin > > On Jan 23, 9:03 am, mdipierro <[email protected]> wrote: > > > can you elaborate on sesssion? > > > On Jan 23, 8:51 am, Robin B <[email protected]> wrote: > > > > This could be another possible shortcut for insert: > > > db.posts[0] = attrs > > > -> db.posts.insert(**attrs) > > > > then a session driver only requires: > > > __getitem__,__setitem__,__delitem__ > > > > Robin > > > > On Jan 23, 8:39 am, Robin B <[email protected]> wrote: > > > > > Session could make use of the new DAL syntax, globals.py: > > > > > rows=db(table.id==record_id).select() > > > > -> row = table[record_id] > > > > > table._db(table.id==record_id).update(**dd) > > > > -> table[record_id] = dd > > > > > Using this syntax session db drivers only need to implement: > > > > > insert(),__getitem__,__setitem__,__delitem__ > > > > > For example MEMDB would not even need a query interface. > > > > > Robin > > > > > On Jan 23, 12:20 am, mdipierro <[email protected]> wrote: > > > > > > I just posted this in trunk following your suggestion. > > > > > > >>> db=SQLDB() > > > > > >>> db.define_table('person',SQLField('name')) > > > > > >>> db.person['name'] > > > > > > <gluon.sql.SQLField object at 0x1327510>>>> > > > > > db.person.insert(name='Robin') > > > > > 1 > > > > > >>> db.person.insert(name='Robin') > > > > > 2 > > > > > >>> db.person[1] > > > > > > <SQLStorage {'update_record': <function <lambda> at 0x13ca930>, 'id': > > > > > 1, 'name': 'Robin'}>>>> db.person[2]=dict(name='Tim') > > > > > >>> db.person[1]=db.person[2] > > > > > >>> db.person[1] > > > > > > <SQLStorage {'update_record': <function <lambda> at 0x13ca930>, 'id': > > > > > 1, 'name': 'Tim'}> > > > > > > >>> del db.person[1] > > > > > > Beautiful, thank you! > > > > > > The lambda stuff requires more thought and work. > > > > > > Massimo > > > > > > On Jan 22, 10:39 pm, Robin B <[email protected]> wrote: > > > > > > > I was just reading UserDict and thought how getitem, setitem, > > > > > > delitem > > > > > > could be added to db.Table for these: > > > > > > > person = db.person[person.id] > > > > > > -> shortcut syntax for db(db.person.id==person.id).select()[0] > > > > > > db.person[person.id] = db.person(age=lambda attrs: attrs.age+=1) > > > > > > -> update a record without having to fetch it first > > > > > > del db.person[person.id] > > > > > > -> delete a record without having to fetch it first > > > > > > > There are more to be had: > > > > > > > len(db.posts) -> count() > > > > > > > for post in db.posts: > > > > > > post.update_record(...) > > > > > > > Robin > > > > > > > On Jan 22, 10:13 pm, mdipierro <[email protected]> wrote: > > > > > > > > Could you elaborate on dictmixin? > > > > > > > > SQLStorage already extends a dict. Which methods do you think are > > > > > > > missing? > > > > > > > > Massimo > > > > > > > > On Jan 22, 9:03 pm, Robin B <[email protected]> wrote: > > > > > > > > > Simple queries could be generated from lists of filters, also > > > > > > > > simple > > > > > > > > queries require fewer ()'s: > > > > > > > > > db(*programmatic_filters).select() > > > > > > > > db(programmatic_filters).select() > > > > > > > > db(db.person.last_name=='Obama',db.person.first_name=='Barack').select > > > > > > > > () > > > > > > > > > vs > > > > > > > > > db((db.person.last_name=='Obama')& > > > > > > > > (db.person.first_name=='Barack')).select() > > > > > > > > > and > > > > > > > > > db(db.person.last_name=='Obama')(db.person.first_name=='Barack').select > > > > > > > > () > > > > > > > > > You have to iteratively build the query from the list, it would > > > > > > > > be > > > > > > > > easier to just pass in the list with or without a * (splat). > > > > > > > > > The real benefit is that drivers that do not allow "|" (GAE, > > > > > > > > MEMDB) > > > > > > > > could use trivial query parsers, which would make those drivers > > > > > > > > simple > > > > > > > > to write and maintain. > > > > > > > > > > > person = db.person(party='Republican') > > > > > > > > > > -> returns a person with default= values populated and > > > > > > > > > > **attrs set > > > > > > > > > > person.first_name = 'John' > > > > > > > > > > response.render(person=person) > > > > > > > > > > -> the record can be printed on a page without inserting it > > > > > > > > > > in db > > > > > > > > > > person.insert() > > > > > > > > > > -> actually inserts the record, returns 1 for success, 0 > > > > > > > > > > for failure, > > > > > > > > > > updates person attrs > > > > > > > > > > Not convinced about this one. > > > > > > > > > The benefit here is that you have easy access to all the default > > > > > > > > values without doing an insert, this become especially useful if > > > > > > > > db.Field(default=lamda:foo(123)) becomes possible. > > > > > > > > > > > person = db.person[person.id] > > > > > > > > > > -> shortcut syntax for > > > > > > > > > > db(db.person.id==person.id).select()[0] > > > > > > > > > > love this one > > > > > > > > > > > db.person[person.id] = db.person(age=lambda attrs: > > > > > > > > > > attrs.age+=1) > > > > > > > > > > -> update a record without having to fetch it first > > > > > > > > > > this depends on the one above I am not convinced about. One > > > > > > > > > could do > > > > > > > > > db.person[person.id].update_record(age=...) > > > > > > > > > db.person[person.id].update_record(age=...) > > > > > > > > -> would require 2 db requests (select,update) > > > > > > > > > db.person[person.id] = db.person(age=...) > > > > > > > > -> would require 1 db request (update) > > > > > > > > > Since __setitem__(self,key,value) maps to self[key] = value, > > > > > > > > if implemented properly in the driver, > > > > > > > > this eliminates the redundant db request. > > > > > > > > > > > del db.person[person.id] > > > > > > > > > > -> delete a record without having to fetch it first > > > > > > > > > > love this one too > > > > > > > > > DictMixin could be used to implement a lot of functionality in > > > > > > > > the > > > > > > > > driver:http://svn.python.org/view/python/trunk/Lib/UserDict.py?view=markup > > > > > > > > > Robin --~--~---------~--~----~------------~-------~--~----~ You received this message because you are subscribed to the Google Groups "web2py Web Framework" group. To post to this group, send email to [email protected] To unsubscribe from this group, send email to [email protected] For more options, visit this group at http://groups.google.com/group/web2py?hl=en -~----------~----~----~----~------~----~------~--~---
