The logging with "app" is only a problem in a context where both string and unicode objects are passed to the logger. In this case it will try to convert the application to a str first (via __str__) and then convert that to unicode. But because there is no __unicode__ method it will assume ascii encoding for the resulting string and fail.
This means that in the places where a single app argument is used its fine and will not crash and its also fine if all strings are of type "str". I mentioned this on irc:
Jun 27 17:15:17 <mvo> tremolux: the only thing that needs invesitgation is if we use a mixed unicode/str in a logger elsewhere that might crash in the same way
but because of the rush to work on the paypal bug I did not put that in the bugreport (I'm sorry for that).
I will revert the parts where there is just a single app argument in the logger and add it to the 5.2 branch
(the added benefit is that the diff also smaller which will make the SRU team happy :)
Thanks for this update.
The logging with "app" is only a problem in a context where both string and unicode objects are passed to the logger. In this case it will try to convert the application to a str first (via __str__) and then convert that to unicode. But because there is no __unicode__ method it will assume ascii encoding for the resulting string and fail.
This means that in the places where a single app argument is used its fine and will not crash and its also fine if all strings are of type "str". I mentioned this on irc:
Jun 27 17:15:17 <mvo> tremolux: the only thing that needs invesitgation is if we use a mixed unicode/str in a logger elsewhere that might crash in the same way
but because of the rush to work on the paypal bug I did not put that in the bugreport (I'm sorry for that).
I will revert the parts where there is just a single app argument in the logger and add it to the 5.2 branch
(the added benefit is that the diff also smaller which will make the SRU team happy :)