Roundup Tracker - Issues


Author ber
Recipients ber, rouilj, tekberg
Date 2013-05-09.14:14:36
Message-id <>
Hi Tom,

good to know that your behaviour is similiar to mine. I was worried a bit about this.

I haven't committed the change, because I'm not confident enough that this is the 
current patch is a good solution.

One thing that worries me is the python method resolution order (mro), because 
class IssueClass(PostgresqlClass, rdbms_common.IssueClass) inherits the 
method  insensitive_like() from both parents. On what basis does the 
PostgresqlClass' method win? Does this work on all circumstances?

Also is this way of using  a method the most elegant way?
Why not using a class variable like it is done with "order_by_null_values"?
And than just check if we have this variable and if so, use the contents?
The dynamics of methods may not be necessary and the default value
would be closer to place in code where it is used. 

If the variable is only defined in case of PostgresqlClass, then there would be no 
need to think about the MRO, thus resolving my first issue.

If you get to it earlier than me, feel free to create a more elegant patch
trying the variable approach.
Date User Action Args
2013-05-09 14:14:36bersetmessageid: <>
2013-05-09 14:14:36bersetrecipients: + ber, rouilj, tekberg
2013-05-09 14:14:36berlinkissue2550805 messages
2013-05-09 14:14:36bercreate