If you want I can make a diff over 1.99.4 and 1.99.5 and try to locate what breaks the old behavior... Also, I understand that it was something experimental, but I think it is pretty usefull for legacy database users.
I will try trunk tomorrow to see if your fix works as expect. Thank you Massimo. Richard On Tue, Aug 28, 2012 at 4:00 PM, Massimo Di Pierro < [email protected]> wrote: > For now I have committed a fix in trunk. I am not sure it belong there? It > is not obvious to me the previous behavior was better or not (the feature > you are using was never documented). > > Pros? Cons? > > > On Tuesday, 28 August 2012 11:56:13 UTC-5, Richard wrote: > >> FYI what you suggest trigger this error message : >> >> <type 'exceptions.SyntaxError'> Object exists and cannot be redefined: id >> >> Richard >> >> On Tue, Aug 28, 2012 at 12:46 PM, Richard Vézina >> <[email protected]>wrote: >> >>> Not sure I follow, in the pass we were allowed to define a legacy id for >>> a table like this : >>> >>> db.define_table('test_**endotoxin', >>> *Field('something_id','id'),* >>> Field('result','decimal(10,2)'**, >>> notnull=True, >>> requires=[IS_NOT_EMPTY(error_**message=T('field can\'t be >>> empty')), >>> IS_DECIMAL_IN_RANGE(0.00,**99999999.99)], >>> required=True >>> ), >>> ...) >>> >>> It still work except at the level of the row object. >>> >>> What you suggest may works, I would have to refactor and create a >>> virtual field for all my tables... >>> I was seeking for a solution that prevent me from refactoring in order >>> to shift from 1.99.4 to 2.0... when there will be a stable version. >>> >>> Richard >>> >>> On Tue, Aug 28, 2012 at 12:39 PM, Massimo Di Pierro < >>> [email protected]> wrote: >>> >>>> How about >>>> >>>> table.id = Field.Virtual(lambda row: row.table.something_id) >>>> >>>> >>>> On Tuesday, 28 August 2012 10:30:37 UTC-5, Richard wrote: >>>>> >>>>> Hello Massimo, >>>>> >>>>> I notice when I try to migrate to 1.99.7 some times ago that I could >>>>> not anymore do something like this : >>>>> >>>>> rows = db(db.table.id>0).select() >>>>> >>>>> for row in rows: >>>>> print row.id >>>>> >>>>> I can did it before 1.99.5. >>>>> >>>>> Has Anthony wrote in this thread : https://groups.google.com/**fo** >>>>> rum/#!msg/web2py/**i7wx6JVTOtw/**FRnScZzhqHgJ<https://groups.google.com/forum/#!msg/web2py/i7wx6JVTOtw/FRnScZzhqHgJ> >>>>> >>>>> We still can do : >>>>> >>>>> db.table.id even if the "real" table id name (or legacy id name) is >>>>> something else like : something_id >>>>> >>>>> To me it is just a matter to have a alias for any legacy id name when >>>>> creating a row object : >>>>> >>>>> Row in 1.99.4 >>>>> <Row {'something_id': 252, 'update_record': <function <lambda> at >>>>> 0x215eaa0>, 'field1': Decimal('5.55'), *'id': 252*, 'delete_record': >>>>> <function <lambda> at 0x215eb18>}> >>>>> >>>>> >>>>> Row in 1.99.5-1.99.7 >>>>> <Row {'something_id': 252, 'update_record': <function <lambda> at >>>>> 0x215eaa0>, 'field1': Decimal('5.55'), 'delete_record': <function <lambda> >>>>> at 0x215eb18>}> >>>>> >>>>> Could we have it back or it is gone? >>>>> >>>>> Thank you. >>>>> >>>>> Richard >>>>> >>>> -- >>>> >>>> >>>> >>>> >>> >>> >> -- > > > > --
