Re: [Traits] dirty trait metadata

Prabhu Ramachandran wrote:
> This automatically builds the shadow _x trait and always calls back 
> _x_changed.  The ShadowProperty can be made smarter so it optionally 
> checks to see if the value changed and only then fire the handler but 
> thats easy to do.  I'll likely do it.  I think this is a common enough 

Now done with docs etc.  I'll check it in for the time being into mayavi 
somewhere (maybe e.mayavi.core.traits).

cheers,
prabhu

from enthought.traits.api import TraitType
from enthought.traits.traits import trait_cast
import numpy

##########################################################################
# `ShadowProperty` trait type.
##########################################################################
class ShadowProperty(TraitType):

    # Not really necessary but specifies the attribute up front.
    trait_type = None

    # Call the notifiers smartly only when the value has really changed.
    # If this is set to False, the notification will always occur.
    smart_notify = True

Re: [Traits] dirty trait metadata

On Sat, Sep 6, 2008 at 00:33, Prabhu Ramachandran
<prabhu@...> wrote:
> Robert Kern wrote:
>
>>> Is there a simple way of doing this?  I know that I could achieve this
>>> if I implemented x as a Property(Array) but is there an easier way?
>>
>> I don't think so. The comparison logic is in C
>> (ctraits.c:setattr_trait), and there are no hooks for this.
>>
>> If you only need coarse-grained notifications, you might consider
>> using an updated Event. This:
>
> OK, so I came up with a new trait type that does this nicely with minimal
> fuss and also does the validation right.

What's the use case again? For the one you demonstrated (notification
that an array's contents have changed), I don't think the
ShadowProperty semantics are particularly the obvious way to do issue
that notification.

--

-- 
Robert Kern

"I have come to believe that the whole world is an enigma, a harmless
enigma that is made terrible by our own mad attempt to interpret it as
though it had an underlying truth."
 -- Umberto Eco

Re: [Traits] dirty trait metadata

Prabhu Ramachandran wrote:
> However, the question about notification on properties remains.  If I 
> were to define something like this:
>
> class Data(HasTraits):
>     x = Property(Float)
>
>     def _set_x(self, val):
>         old = self._x
>         self._x = val
>         self.trait_property_changed('x', old, val)
>
> d = Data()
>
> Then even if someone did d.set(x=1, trait_change_notify=False), you'd 
> get a notification that x changed, right?  My patch "fixed" that perhaps 
> incorrectly but the solution was clean and solved the above issue.  I'd 
> like to know your thoughts on this.
>
>   

Yes, I have not forgotten this, although it would serve you better if 
you wrote up a ticket (and attached your patch to that as well). I know 
you have trouble creating tickets, but it does make it more trackable. 
If you do, please put it in the Traits 3.0.3 milestone, since the 3.0.2 
milestone is pretty much in the bag.

I understand what you want to do, and I believe the core of your 
solution is valid. The only problem I have is folding it into the 
solution we already have (and which your example is not using).