Re: documenting widgets was: release 0.4.1 bug triage

I don't think the templateCssPath has anything to do with whether 
templateCssPath is declared in the initializer or the prototype.  After 
all, we have lots of widgets that have string parameters and they work 
fine.  Obviously, loading two CSS files that define conflicting rules 
for .dojoToolbar is no good, but that's an unrelated issue.

Like I said in my last mail, show me some examples of these subtle 
problems of which you speak.  I stress the word *example*.


Tom Trenka wrote:
> The templatePath/templateCssPath issue I'm referring to is the one where 
> you can "override" the default value set in a widget by adding that 
> attribute to a widget's markup:
>
> <div dojoType="someWidget" templateCssPath="../../path/to/css.css"></div>
>
> ...we've known for a long time that doing this will cause *all* widgets 
> of type someWidget to use the new CSS template as opposed to the one 
> defined in the widget code (you, me and Dustin talked about it briefly 
> when we started talking about theming solutions).    The reason for this 
> is because it is a shared property--because it's been defined on the 
> prototype of the widget and not the widget itself.
>
> You can declare widget parameters within extend just fine--if you 
> understand the consequences of that definition.  A good example would be 
> the declaration of values that shouldn't be touched by a user of the 
> widget at all, such as isContainer or widgetType.  But full 
> understanding of *exactly* what is going on when doing that is really 
> needed.  I'm not convinced that's the case; as you've mentioned, it's 
> less full understanding and more "this is a convention that Alex started".
>
>     Hmm, well if it doesn't work then we have to change it.  As usual, I
>     don't care much either way but I want to have a standard that everyone
>     follows.  (Here an inconsistency is especially bad since things declared
>     in a base class's initializer will overwrite things declared in the
>     subclass's prototype.)
>
>
> It's not as much "overwrite" as much as it is "masks"; minor point but 
> we're about to get back into the whole language terminology argument, so 
> I'll leave that for a different thread.  Either way I agree, we should 
> have a standard.  I also think this change should be a 0.5 one and not a 
> 0.4.1 one, so...I suggest we at least hold off until after any kind of 
> sub-point release is made (or branch, for that matter).
>
> trt
>
>
> ------------------------------------------------------------------------
>
> _______________________________________________
> dojo-contributors mailing list
> dojo-contributors@xxxxxxxxxxxxxxx
> http://dojotoolkit.org/mailman/listinfo/dojo-contributors