> > 2) Going from 'private' to 'protected' because of the need to 
> > subclass in a particular proprietary application should never be a 
> > problem. This strenghtens the API.
> 
> This is not always the case. See 
> http://opensimulator.org/mantis/view.php?id=3072. Sometimes there are good 
> reasons 
> for keeping methods private.


I think that discussion has to be separated; I don't know the reason for those 
members to be kept private (other than that the plugin 'wasn't intended to be 
subclassed' which is fair enough) but I can say that generally the only reason 
to hesitate over going from 'private' to 'protected' is if code is written so 
that state integrity is pivotal for method execution - something that usually 
points to the code being brittle in the first place. (Compare with calling 
virtual members from constructors for a similar concern)

 

I am perfectly aware that this is touching on religion so I won't push the 
issue further; suffice to say that "in most cases" we should not have a problem 
with it.

 

For the interested, I suggest you google for religious flame wars on the proper 
use of "internal" and "sealed" as well. ;-)

 

By the way, I take your +1's as meaning we're discussing this in order to form 
an explicit concensus - if I have recieved no strong objections, I will add 
this to the code conventions by friday 20/2 2009.

 

Best Regards,

/Stefan

 
_______________________________________________
Opensim-dev mailing list
[email protected]
https://lists.berlios.de/mailman/listinfo/opensim-dev

Reply via email to