working on tracing error gets me to line 3196 of sql.py:
sql_o += ' ORDER BY %s' % ', '.join(['%s.%s'%(t,x) for t in
w2p_tablenames for x in ((hasattr(self._db[t],'_primarykey') and
self._db[t]._primarykey) or ['id'])])
Problem is that with this table definition:
db.define_table('mytable',
Field('my_id', 'id')
)
This line:
hasattr(db.mytable, '_primarykey') # returns False
It seems like the above line should return True and
db.mytable._primarykey should equal 'my_id'. Which makes me think the
problem actually goes back to define_table.
-Mike
On Sep 21, 4:44 pm, mwolfe02 <michael.joseph.wo...@gmail.com> wrote:
> I'm using a legacy tables and trying to use the shortcut method of
> returning a row/record by passing the value of the ID field directly
> to the table. It appears that 'id' is still hardcoded into the logic
> at some level, though. This works:
>
> db.mytable(db.mytable.my_id==1)
>
> But this does not:
>
> db.mytable(1)
>
> # returns ProgrammingError: ('42S22', "[42S22] [Microsoft][ODBC SQL
> Server Driver][SQL Server]Invalid column name 'id'. (207)
> (SQLExecDirectW)")
> # for a table defined as follows:
>
> db.define_table('mytable',
> Field('my_id', 'id')
> )
>
> db.mytable(1)
>
> # returns KeyError: 'id'
> # for a table defined as follows:
>
> db.define_table('mytable',
> Field('my_id', 'id'),
> primarykey=['my_id']
> )
>
> I can understand the second case failing, as primarykey seems like it
> would be usually used to define a multi-field key. In such a case, a
> single value would not be enough to identify a record, anyway. I
> tried this simply as a workaround for the original problem, but with
> no success.
