[dojo-contributors] Dojo 1.4.2 RC build this Monday

Ben Hockey neonstalwart at gmail.com
Wed Mar 3 13:21:42 EST 2010


not sure if i'll be at the meeting tonight but i added an agenda item  
to discuss http://bugs.dojotoolkit.org/ticket/10788 since it's tagged  
for 1.4.2 and is one of possibly a few tickets which might be holding  
up the release.  in case i'm not at the meeting, i'll outline my  
thoughts here.

i don't mind if the fix for this makes it into 1.4.2 or not but i do  
feel strongly that it should be included in a release as soon  
practical once it's available.

to summarize the problem, in some cases, making a change to the  
prototype of a superclass, whether via changing Foo.prototype directly  
or via Foo.extend, it is not being inherited by the subclasses.  i  
have made a page which demonstrates some of the cases where the change  
is not inherited http://neonstalwart.dojotoolkit.org/test/declare.html  
- all of the examples here use Foo.extend to change the prototype.

from what i understand (and eugene can correct me if i'm wrong), the  
constructor contains some meta data which helps this.inherited fast  
track directly to the next function in the inheritance chain rather  
than looking at each prototype in the inheritance chain.  the issue is  
caused when one of the prototypes changes and the 'fast-track'  
function is cached as the original function.  this happens in 2  
general cases i found.  in both cases D inherits from A, B and C.  A  
and B have function foo and C does not.  in this scenario B's foo will  
be cached in the constructor meta data.  the 2 cases which showed  
problems for me were:
  case 1. if B's function foo is changed in the prototype, D will  
still look to the B's original foo
  case 2. if C's prototype changes and now includes a foo function, D  
will still look to B's original foo

i feel that while this remains the case we are vulnerable to some  
harsh criticism and in my opinion deservedly so.  this is why i think  
the fix should be included as soon as possible.

if i'm at the meeting tonight, i probably won't have much more to add  
to this and in my opinion eugene is certainly the authority to talk  
more about the details of this and provide the solution to this  
problem.  i just wanted to try and outline the problem as i saw it so  
others can form their own opinions about the severity and decide if  
1.4.2 needs to wait for this or not.

thanks,

ben...

On Feb 24, 2010, at 8:07 PM, James Burke wrote:

> Just a heads up that we will be doing the Dojo 1.4.2 RC build this
> coming Monday, March 1, with the final build happening likely that
> Monday after that, March 8.
>
> James
> _______________________________________________
> dojo-contributors mailing list
> dojo-contributors at mail.dojotoolkit.org
> http://mail.dojotoolkit.org/mailman/listinfo/dojo-contributors



More information about the dojo-contributors mailing list