Menu
▾
▴
Re: [Dynapi-Dev] Fixes n stuff
|
From: Scott A. L. <sc...@sc...> - 2000-10-30 22:20:36
|
I don't know how it happened, but none of those array modifications in
deleteChild were in the patch I posted to SourceForge. Pascal is
correct that deleteChild already takes care of handling the assigned and
unassigned arrays.
The children[] array was removed because of the inheritance problem with
the WidgetX model: the children[] array was being shared. So children
is now created when you first add children to an object.
Pascal's proposed fix to the WidgetX model (using this.superClass, see
earlier posts) should correct this, but it hasn't been accepted yet as
part of the standard model for widgets.
I played around with the proposed fix for some time yesterday, and it
seems to work:
function Widget(){
this.superClass = DynLayer <-- defines the superClass constructor as a
method
this.superClass() <-- invokes the superClass constructor, ensuring
that this object is a unique instance
...
}
Widget.prototype = new DynLayer() <--- Widget prototype is based on a
unique instance of DynLayer
If this were adopted, we could go ahead and put children[] back into the
constructor of DynLayer and DynDocument and not need the hasChildren
boolean. this.superClass() would make sure that each Widget instance
has unique values.
I think this is a fairly straightforward compromise, and you only have
to add two short lines. Since everyone seems to like the simplicity of
the WidgetX model, I think this is the way to go. Of course, if there
were a way to shorten it to one line...
scottandrew
Pascal Bestebroer wrote:
>
> Ok, here are a few possible fixes for the latest release:
>
> in: DynLayer.prototype.removeChild()
> change this: DynLayer.unassigned[DynLayer.unAssigned.length]=child
> into this: DynLayer.unassigned[DynLayer.unassigned.length]=child
> error was: the unAssigned array is not with a capital A ;)
>
> deleteElement contains a MAJOR bug:
>
> DynLayer.prototype.deleteElement=function() {
> this.created=false
> /*
> DONT DO THE FOLLOWING!
> It's already being done by the functions that call deleteElement, so don't
> do it here again - major problems!
>
> DynAPI.removeFromArray(this.dyndoc.all,this,true)
> delete this.dyndoc.allID[this.id]
> DynLayer.unassigned[DynLayer.unassigned.length]=this
> DynLayer.unassignedID[this.id]=this
> */
> if (is.ns4) {
> this.elm.releaseEvents(Event.LOAD)
> this.elm.visibility="hide"
> } else {
> this.elm.style.visibility="hidden"
> this.elm.innerHTML=""
> this.elm.outerHTML=""
> // + following line won't work on IE4, so check for IE5
> if (is.ie5 && this.elm.children.length>0) this.elm.removeNode(true)
> }
> this.elm={}
> this.doc={}
> this.css={}
> this.deleteChildElements()
> }
>
> Also the inline code is totally broken, I've got a working version
> but it depends on the children array still "alive". I don't know why
> the children[] is removed (or atleast not an array at the beginning) this
> makes coding a tricky thing (always having to check for the children array
> being an array)
>
> the widget modifications posted a few days ago should make the children[]
> problems solved,
> so that the array can always be an array, so that checks don't have to be
> done maybe saving
> a few code-breaking situations, speed and file-sizes..
>
> anyway, here's the correct inline code (might need a children[] fix as
> mentioned, or preferably
> restore the children array to an array) :
>
> /*
> Core DynAPI Distribution
> AddOn extensions
> */
> DynAPI.findLayers=function(dyndoc,or) {
> var divs=[]
> or=or||dyndoc
> if (is.ns4) divs=dyndoc.doc.layers
> if (is.ns5) divs=dyndoc.doc.getElementsByTagName("DIV")
> if (is.ie) divs=dyndoc.doc.all.tags("DIV")
> for (var i=0; i<divs.length; i++) {
> if(DynAPI.isDirectChildOf(divs[i],dyndoc.elm)) {
> var id=is.ns4? divs[i].name : divs[i].id
> var dlyr=new DynLayer(id)
>
> dlyr.parent=dyndoc
> dlyr.assignElement(divs[i])
> dlyr.dyndoc=or
>
> // new line here, not sure why I did it, but it fixed a few things
> // for events I think.
> if (dyndoc.getClass()!=DynDocument) dlyr.isChild=true
>
> if (or.getClass()!=DynDocument) dlyr.isChild=true
> else {
> // this part was needing some fixing for Scott's new code:
> DynAPI.removeFromArray(DynLayer.unassigned, dlyr, true)
> delete DynLayer.unassignedID[dlyr.id]
>
> or.all[or.all.length]=dlyr
> or.all[dlyr.id]=dlyr
> or.allID[dlyr.id]=dlyr
> }
>
> dyndoc.children[dyndoc.children.length]=dlyr
>
> var index=id.indexOf("Div")
> if (index>0) dyndoc.doc.window[id.substr(0,index)] = dyndoc.all[id]
>
> if (is.ns4) {for (ict in dlyr.doc.images)
> dlyr.doc.images[ict].lyrobj=dlyr}
> else if (is.ns5) {for (ict in dlyr.doc.images)
> dlyr.doc.images[ict].lyrobj=dlyr.elm}
> else {for (ict in dlyr.elm.all.tags("img"))
> dlyr.elm.all.tags("img")[ict].lyrobj=dlyr}
>
> if (dlyr.updateValues) dlyr.updateValues()
>
> DynAPI.findLayers(dlyr,or)
> }
> }
> }
> DynLayer.prototype.updateValues=function() {
> if (is.ns) {
> this.x=parseInt(this.css.left)
> this.y=parseInt(this.css.top)
> this.w=is.ns4? this.css.clip.width : parseInt(this.css.width)
> this.h=is.ns4? this.css.clip.height : parseInt(this.css.height)
> if (is.ns4)
> this.clip=[this.css.clip.top,this.css.clip.right,this.css.clip.bottom,this.c
> ss.clip.left]
> this.bgColor = this.doc.bgColor!="this.doc.bgColor"?this.doc.bgColor:null
> this.bgImage = this.elm.background.src!=""?this.elm.background.src:null
> this.html = this.innerHTML = this.elm.innerHTML = ""
> }
> else if (is.ie) {
> this.x=this.elm.offsetLeft
> this.y=this.elm.offsetTop
> this.w=is.ie4? this.css.pixelWidth : this.elm.offsetWidth
> this.h=is.ie4? this.css.pixelHeight : this.elm.offsetHeight
> this.bgImage = this.css.backgroundImage
> this.bgColor = this.css.backgroundColor
> this.html = this.innerHTML = this.elm.innerHTML
> }
> this.z = this.css.zIndex
> var b = this.css.visibility
> this.visible = (b=="inherit"||b=="show"||b=="visible"||b=="")
> }
> DynAPI.getModel=function() {
> dom='DYNAPI OBJECT MODEL:\n\n+DynAPI\n'
> for (var i=0; i<DynAPI.documents.length; i++) {
> dom+=' +'+DynAPI.documents[i].toString()+'\n'
> for (var j=0; j<DynAPI.documents[i].all.length; j++)
> dom+='+'+DynAPI.documents[i].all[j].toString()+'\n'
> }
> alert(dom)
> }
> DynAPI.isDirectChildOf = function(l, parent) {
> if(is.ns) return (l.parentLayer == parent)
> for(var p=l.parentElement;p;p=p.parentElement)
> if(p.tagName.toLowerCase()=='div') return p==parent
> return !parent.tagName
> }
>
> Some doubts about the event code in the current release, there is some code
> for the
> old rtmouseup and rtmousedown events.. don't think they are needed (at least
> not in that
> state) I've got the following solution working:
>
> Use all of Scott's this.button changes, and change the handleEvent()
> method into this:
>
> EventListener.prototype.handleEvent=function(type,e)
>
> if (type=='mousedown' || type=='mouseup') {
> if (e.button==2) type='md'+type
> if (e.button==3) type='rt'+type
> }
> if (this["on"+type]) this["on"+type](e)
> }
>
> should do correct event bubbling, and still make different events for
> rtmouseup/rtmousedown
> and mdmouseup/mdmousedown (middle button)
>
> that's it for now.. have fun :)
>
> Pascal Bestebroer
> pa...@dy...
> http://www.dynamic-core.net
>
> _______________________________________________
> Dynapi-Dev mailing list
> Dyn...@li...
> http://lists.sourceforge.net/mailman/listinfo/dynapi-dev
|
