> On Saturday, October 11, 2008 at 2:50AM Per Cederberg wrote: > > > On Thu, Oct 9, 2008 at 6:50 PM, Jason Bunting > <[EMAIL PROTECTED]> wrote: > >> Well... I think your case here is pretty uncommon. This is because the > >> __dom__() function is really supposed to create a *new* DOM node. > >> Otherwise people might run into issues when adding an object twice > >> into the DOM tree. > > > > Excuse my ignorance, and permit me to ask a few questions so I can > explore > > this further... > > > > Line item 6 in the "DOM Coercion Rules," as posted in the documentation, > > states: > > > > 6. Objects that have a .dom(node) or .__dom__(node) > > method are called with the parent node and their > > result is coerced using these rules. > > > > So, perhaps there is some confusion because of the documentation, but I > > don't see how my example code violates anything. > > There is no specification on this, it is just kind of what you'd > expect. Why would otherwise the parent node be an input parameter? If > the result is constant, no parameter in the world can change that.
I could just as easily ask why the result of the call would be coerced to a DOM element - and I can think of situations where knowing about the intended parent node would be useful to a 'widget' when its __dom__ or dom function is called with it. Consider, for example, that a widget might want to use the dimensions of the parent node for setting its behavior - when __dom__ is called, I can gather information about the passed-in parent node and update my widget's size or behavior before passing back the thing that will be coerced for placement in the DOM. > > I am confused by your statement that "Otherwise people might run into > issues > > when adding an object twice into the DOM tree" - using my example, if > > someone were to try to add myWidgetInstance to the DOM twice, the > behavior > > would be exactly as I would expect it - it is the same instances, and > thus > > it would only appear once (because the call to __dom__ would return the > same > > instance). If the developer doesn't understand that this would happen, > then > > they have other problems. Unless they instantiate another instance, > there > > should only be one. > > I wasn't thinking about widgets, but rather situations were you'd > added a dom() method to various other objects. For convenience. I won't fault you for your particular view of how one might use certain facilities of MochiKit if you don't fault me for mine. :) > >> But sure, there is an inconsistency here. My suggestion would be to > >> just work around it instead: > >> > >> removeElement(myWidgetInstance.widgetDomRepresentation); > > > > IMO, that's terrible. It breaks encapsulation because now something that > > should be private is made explicitly public. I don't want a workaround, > I > > want consistency in MochiKit's API. > > I shouldn't start an OO discussion here, but in my opinion the fields > in an object are all public unless names are prefixed with an "_". In the widgets I develop, the private fields are not accessible because of closures. Doesn't make sense to me to call something private when it isn't really private (or when it is only considered private because of a naming convention). > > I appreciate your comments, and while an API for widget building may > provide > > some useful help, it isn't what I am looking for at the moment. The way > I > > have built widgets up to now (successfully, and for quite a while) is > pretty > > much the way my example implies. It works beautifully and is simple > enough > > to be understood without an entire widget framework (notwithstanding the > > fact that some help from using one might eventually be better than my > > approach). I would simply like some consistency in the API - the > following > > functions all use the DOM Coercion Rules: > > > > appendChildNodes > > insertSiblingNodesBefore > > insertSiblingNodesAfter > > createDOM > > replaceChildNodes > > ... > > > > If those do, so should any of the others that expect DOM elements: > > > > removeElement > > swapDOM > > ... > > Ehm... The proposed MochiKit.Widget isn't "an entire widget > framework". I just pointed at it for example, not to force you to > change your code or your ways. It is far more of a widget framework than what I currently use, so to me it is "an entire widget framework." :) Besides, it is meant to do just that, isn't it? I think the word "entire" is rather subjective. And I realize you were not trying to force me into anything. > I don't oppose changing there MochiKit.DOM functions, I'm just of the > opinion that it isn't much of a problem. And if it is, I'd suggest > that we check typeof(o.dom) == "object" or something. So that we know > for sure that what is being removed is an existing DOM node, not > something that was created by our call to o.dom()... Also, doing that > would increase our compability with Dojo et al. I will create a patch and it can go up for discussion - I already modified removeElement() a week ago to test things out and all I did was simply call coerceToDom() within removeElement(), just as most of the other functions that use it do so. I will make sure I look at all of the tests, try to provide others, etc. Jason Bunting --~--~---------~--~----~------------~-------~--~----~ You received this message because you are subscribed to the Google Groups "MochiKit" group. To post to this group, send email to [email protected] To unsubscribe from this group, send email to [EMAIL PROTECTED] For more options, visit this group at http://groups.google.com/group/mochikit?hl=en -~----------~----~----~----~------~----~------~--~---
