I finally found the solution using COALESCE to merge and order date columns:
exams_mutex = db().select(db.dataset1.date.coalesce(db.dataset2.date), db. dataset1.ALL, db.dataset2.ALL, left=[db.dataset1.on(db.mutex.i==0),db. dataset2.on(db.mutex.i==1)], orderby=db.dataset1.date.coalesce(db.dataset2. date)) Mamisoa Le jeudi 20 décembre 2012 16:08:34 UTC+1, Massimo Di Pierro a écrit : > > Sorry I do not have a good solution at the moment. Can you open a ticket > about this so we'll will not forget. For now you may want to consider > creating a database view and selecting from the view. You would need to > create a model (readonly) to access the view. > > On Wednesday, 19 December 2012 16:19:33 UTC-6, Mamisoa Andriantafika wrote: >> >> Best results to UNION I could have was using the mutex table trick: >> >> http://www.xaprb.com/blog/2006/05/26/how-to-write-full-outer-join-in-mysql/ >> with this syntax: >> >> exams_mutex = db().select(db.dataset1.ALL, db.dataset2.ALL, >> left=[db.dataset1.on(db.mutex.i==0),db.dataset2.on(db.mutex.i==1)]) >> >> Two problems left: >> >> 1- how to merge the 2 date columns >> 2- how to order then by date >> >> SQL answer is there: >> >> http://stackoverflow.com/questions/8245630/mysql-combine-two-date-fields-and-then-order-by >> >> SELECT >> [some info], >> GREATEST( ticket_date, ticket_history_date ) as latest_date >> FROM >> [tables and join] >> ORDER BY >> latest_date >> >> >> Is is possible using to process it using DAL? >> >> Le dimanche 16 décembre 2012 17:40:22 UTC+1, Mamisoa Andriantafika a >> écrit : >>> >>> Sorry I still get: "Cannot | incompatible Rows objects". >>> >>> I'll change the field name "date" you are very right. >>> >>> Le dimanche 16 décembre 2012 15:49:57 UTC+1, Massimo Di Pierro a écrit : >>>> >>>> My bad. Thry this: >>>> >>>> fields1 = [db.dataset1.date, db.dataset1.param1, db.dateset1.patient_id] >>>> fields2 = [db.dataset2.date, db.dataset2.test1, db.dateset2.patient_id] >>>> rows = ( db(db.dataset1).select(*fields1) | >>>> db(db.dataset2).select(*fields2) ).sort(lambda row: row.date) >>>> >>>> Mind that having a column called "date" will result in major headaches >>>> in the future. >>>> >>>> On Sunday, 16 December 2012 03:15:46 UTC-6, Mamisoa Andriantafika wrote: >>>>> >>>>> Hi again, >>>>> >>>>> "|" seems not to work because the column numbers is different between >>>>> the 2 tables? >>>>> >>>>> >>>>> Le samedi 15 décembre 2012 20:23:44 UTC+1, Massimo Di Pierro a écrit : >>>>>> >>>>>> If you have lots of records you may be able to do it with a database >>>>>> view but that may be db specific. >>>>>> >>>>>> If you don't have too many records you can do: >>>>>> >>>>>> rows = ( db(db.dataset1).select() | db(db.dataset2).select() >>>>>> ).sort(lambda row: row.date) >>>>>> >>>>>> >>>>>> >>>>>> On Saturday, 15 December 2012 08:43:35 UTC-6, Mamisoa Andriantafika >>>>>> wrote: >>>>>>> >>>>>>> Hi, >>>>>>> >>>>>>> I have this db model: >>>>>>> >>>>>>> db.define_table('patients', >>>>>>> Field('name', 'string', length=32), >>>>>>> Field('firstname', 'string', length=32), >>>>>>> Field('dob', 'date'), >>>>>>> format='%(name)s') >>>>>>> >>>>>>> db.define_table('dataset1', >>>>>>> Field('date', 'date', length=32), >>>>>>> Field('param1', 'string', length=50), >>>>>>> Field('param2', 'string', length=50), >>>>>>> Field('patient_id', db.patients, writable=False, readable=False)) >>>>>>> >>>>>>> db.define_table('dataset2', >>>>>>> Field('date', 'date', notnull=True), >>>>>>> Field('test1', 'text'), >>>>>>> Field('patient_id', db.patients, writable=False, readable=True)) >>>>>>> >>>>>>> I'd like to show in one view, for 1 patient_id, all the >>>>>>> corresponding dataset1 and dataset2 ordered by date. >>>>>>> >>>>>>> What query should I use? Do I have to use an intermediate table >>>>>>> 'history' to record each activity in dataset1/2 to get a result? >>>>>>> >>>>>>> Thanks for help. >>>>>>> >>>>>> --