dynapi-dev

[Dynapi-Dev] Next Generation

[Eytan H. <ey...@tr...>]

Hi,
Since recent discussion has led to the reorganization of the code, and
basically rewriting it I wanted to know what you guys think about a serious
redo and not only a face lift. Then (yes I know I'm bugging you about this
again but hey maybe someone will get convinced) we can at least implement
some of the real OO code. Is there any way I can convince some of you. There
are many reasons I want to do this but mainly stability. And also changing
the amount of DynLayers used for everything and the weight of dynlayers. I
have an example (will try to get it on a public server so you can see the
bug too) that when you add more and more dynlayers suddenly some of them
start to disappear and then appear out of the blue.

all rocks will be accepted,
8an

Re: [Dynapi-Dev] Next Generation

[Dann <da...@to...>]

Hi Eytan,

As you know, I am interested. I would also like to see this project maturing
with respect to the insight into some of the other problems that keep this
effort from becoming a true reference in the field, it has tremendous value but
it falls short of expectations :
1. It fails to be cross-platform/cross-browser because there are no levels of
functionality identified. Instead it's a single-shot full whistles and bells
API, of which 'this' doesn't work on Mozilla, that doesn't work on IE5 for
Mac... etc...
2. It fails to be stable because some in the dev community find their little
sinewave animation bull* more important than stabilizing DynLayer itself. It's
like some prefer a funny racoon tail hanging from their rearview mirror, instead
of a decent running engine.

At least I detracted some of the stones coming your way - they're now heading my
way :))

CU,
 Dann



Eytan Heidingsfeld wrote:

> Hi,
> Since recent discussion has led to the reorganization of the code, and
> basically rewriting it I wanted to know what you guys think about a serious
> redo and not only a face lift. Then (yes I know I'm bugging you about this
> again but hey maybe someone will get convinced) we can at least implement
> some of the real OO code. Is there any way I can convince some of you. There
> are many reasons I want to do this but mainly stability. And also changing
> the amount of DynLayers used for everything and the weight of dynlayers. I
> have an example (will try to get it on a public server so you can see the
> bug too) that when you add more and more dynlayers suddenly some of them
> start to disappear and then appear out of the blue.
>
> all rocks will be accepted,
> 8an
>
> _______________________________________________
> Dynapi-Dev mailing list
> Dyn...@li...
> http://lists.sourceforge.net/lists/listinfo/dynapi-dev

RE: [Dynapi-Dev] Next Generation

[Eytan H. <ey...@tr...>]

I owe you for the stones bit.
;-)

RE: [Dynapi-Dev] Next Generation

[Matthew A. S. <ms...@go...>]

Very well said.  The foundation of this project the DynLayer is not stable
yet we just keep piling more and more onto it...

M.

-----Original Message-----
From: dyn...@li...
[mailto:dyn...@li...]On Behalf Of Dann
Sent: Sunday, January 28, 2001 5:54 AM
To: dyn...@li...
Subject: Re: [Dynapi-Dev] Next Generation


Hi Eytan,

As you know, I am interested. I would also like to see this project maturing
with respect to the insight into some of the other problems that keep this
effort from becoming a true reference in the field, it has tremendous value
but
it falls short of expectations :
1. It fails to be cross-platform/cross-browser because there are no levels
of
functionality identified. Instead it's a single-shot full whistles and bells
API, of which 'this' doesn't work on Mozilla, that doesn't work on IE5 for
Mac... etc...
2. It fails to be stable because some in the dev community find their little
sinewave animation bull* more important than stabilizing DynLayer itself.
It's
like some prefer a funny racoon tail hanging from their rearview mirror,
instead
of a decent running engine.

At least I detracted some of the stones coming your way - they're now
heading my
way :))

CU,
 Dann



Eytan Heidingsfeld wrote:

> Hi,
> Since recent discussion has led to the reorganization of the code, and
> basically rewriting it I wanted to know what you guys think about a
serious
> redo and not only a face lift. Then (yes I know I'm bugging you about this
> again but hey maybe someone will get convinced) we can at least implement
> some of the real OO code. Is there any way I can convince some of you.
There
> are many reasons I want to do this but mainly stability. And also changing
> the amount of DynLayers used for everything and the weight of dynlayers. I
> have an example (will try to get it on a public server so you can see the
> bug too) that when you add more and more dynlayers suddenly some of them
> start to disappear and then appear out of the blue.
>
> all rocks will be accepted,
> 8an
>
> _______________________________________________
> Dynapi-Dev mailing list
> Dyn...@li...
> http://lists.sourceforge.net/lists/listinfo/dynapi-dev

Re: [Dynapi-Dev] Next Generation

[Robert R. <rra...@ya...>]

I would love to see a more modular approach to the DynAPI as well.
For instance I like Pascal's DynObject for which DynLayer and
DynDocument belong since they share common functionality.  The same
could also be done for events (mouse and keyevents, etc all have some
common functionality).  I think the much of the dynlayer/dyndocument
stuff could be moved out of the DynAPI.js so that the DynAPI can be
used to support more than just dynlayers.

I don't really like the idea of OO in JS. The problem is Javascript
does not support OO. The DynAPI as it is now uses an Object-oriented
approach. There is no standard for trying to make javascript use OO
and their are lots of problems trying to do this in JS. As it stands
now, the widgets use proper subclassing, atleast as close as possible
in Javascript. I don't see why the next gener. couldn't use the same
approach. Its not really and OO issue, but a DynAPI design issue.

Maybe people can make some proposals to the list as to how to design
the next version.

However, the current version still needs to be improved with
Mozilla/NS 6 support (as well as other fixes) before any thoughts of
starting the next version begins.


-- 
// Robert Rainwater

On 1/28/2001, 8:23:31 AM EST, Eytan wrote about "[Dynapi-Dev] Next Generation":

> Hi,
> Since recent discussion has led to the reorganization of the code, and
> basically rewriting it I wanted to know what you guys think about a serious
> redo and not only a face lift. Then (yes I know I'm bugging you about this
> again but hey maybe someone will get convinced) we can at least implement
> some of the real OO code. Is there any way I can convince some of you. There
> are many reasons I want to do this but mainly stability. And also changing
> the amount of DynLayers used for everything and the weight of dynlayers. I
> have an example (will try to get it on a public server so you can see the
> bug too) that when you add more and more dynlayers suddenly some of them
> start to disappear and then appear out of the blue.

> all rocks will be accepted,
> 8an


> _______________________________________________
> Dynapi-Dev mailing list
> Dyn...@li...
> http://lists.sourceforge.net/lists/listinfo/dynapi-dev


----------------------
DynAPI Snapshots: http://dynapi.sourceforge.net/snapshot/
DynAPI Homepage: http://dynapi.sourceforge.net/

Re: [Dynapi-Dev] Next Generation

[Dann <da...@to...>]

Hi,

I have my doubts about the power of ECMAscript as well, but the practice of doing
OO in prototype-based languages (like ECMAscript) is documented, albeit not in the
particular case of ECMAscript itself. I'll take a closer look on the subject to see
what I can dig up...

CU,
 Dann

RE: [Dynapi-Dev] Next Generation

[Eytan H. <ey...@tr...>]

OO in JS or ECMA script is documented. If you read carefully you will notice
that JS was designed as a Class Oriented language that is basically the same
as OO (Objects and Classes are basically the same). The only difference
between OO and CO is the way it does inheritance hence my new inherit
function.

About the NS6 I think that can be worked on at the same time while the NG is
being worked on. One of the reasons of designing the NG like this is so that
we can update the Core of DynAPI without having to change everything just
one file (dynlayer.js).
8an

Re: [Dynapi-Dev] Next Generation

[Dann <da...@to...>]

Hold on there,

... take care with the terms you use ! That's not how I understood it :

the OO philosophy embraces both class-based (CO) and prototype-based (PO)
objects. It's not OO versus CO, it's CO versus PO ! And if I read the specs
correctly, JS is supposed to be a prototype-based language, not a class-based
language.

CU,
 Dann

Eytan Heidingsfeld wrote:

> OO in JS or ECMA script is documented. If you read carefully you will notice
> that JS was designed as a Class Oriented language that is basically the same
> as OO (Objects and Classes are basically the same). The only difference
> between OO and CO is the way it does inheritance hence my new inherit
> function.
>
> About the NS6 I think that can be worked on at the same time while the NG is
> being worked on. One of the reasons of designing the NG like this is so that
> we can update the Core of DynAPI without having to change everything just
> one file (dynlayer.js).
> 8an
>
> _______________________________________________
> Dynapi-Dev mailing list
> Dyn...@li...
> http://lists.sourceforge.net/lists/listinfo/dynapi-dev

RE: [Dynapi-Dev] Next Generation

[Eytan H. <ey...@tr...>]

Lets really get the terms straight.

Class-Oriented: You develop classes. You do this be using the prototype of
an object. Then this object becomes basically a prototype, a class and then
for inheritance an object says use this objects prototype and add these
members.
Object-Oriented: You develop classes by defining an interface and then
implementing this interface. To use this object you create a new object
based on this class. Inheritance is dealt with by saying I want to inherit
this and this class. Then your new object has the same members as those
classes with pointers to their implementation.

There is no PO only CO and OO the difference is inheritance and memory
management but since there is no memory management in JS there is no
difference but inheritance.
8an

Re: [Dynapi-Dev] Next Generation

[Dann <da...@to...>]

Hi 8tan,

You are looking for the stones from the computer science theory people too,
aren't you ? LOL !

No really, where the hell do you get this stuff from ? It sounds like
revisionist CS theory to me :))

CU,
 Dann



Eytan Heidingsfeld wrote:

> Lets really get the terms straight.
>
> Class-Oriented: You develop classes. You do this be using the prototype of
> an object. Then this object becomes basically a prototype, a class and then
> for inheritance an object says use this objects prototype and add these
> members.
> Object-Oriented: You develop classes by defining an interface and then
> implementing this interface. To use this object you create a new object
> based on this class. Inheritance is dealt with by saying I want to inherit
> this and this class. Then your new object has the same members as those
> classes with pointers to their implementation.
>
> There is no PO only CO and OO the difference is inheritance and memory
> management but since there is no memory management in JS there is no
> difference but inheritance.
> 8an
>
> _______________________________________________
> Dynapi-Dev mailing list
> Dyn...@li...
> http://lists.sourceforge.net/lists/listinfo/dynapi-dev

Re: [Dynapi-Dev] Next Generation

[Dann <da...@to...>]

Hi 8tan,

First let's get some common ground through this article, note the use of
'-based' and '-oriented' :

http://www.mactech.com/articles/frameworks/7_6/Prototype-based_OOLs_Evins.html

and then come to terms with the fact that the terms prototype-based and
object-based (<>object-oriented !!!) are very close :

http://www.cpsc.ucalgary.ca/~gao/60903/60903021.htm

Were we talking about the same thing all along ? The irony :)) LOL !

CU,
 Dann

Re: [Dynapi-Dev] Next Generation

[Dann <da...@to...>]

This is also a must read, me thinks :

http://www.urc.bl.ac.yu/manuals/jsobj/index.html

RE: [Dynapi-Dev] Next Generation

[Cameron H. <ca...@bi...>]

netscape themselves refer both javascript and java as object-based
languages. javascript they say is prototype-based, and java is class-based.
both are object-based.

straight from the horses mouth
http://developer.netscape.com/docs/manuals/js/client/jsguide/obj2.htm

> -----Original Message-----
> From: dyn...@li...
> [mailto:dyn...@li...]On Behalf Of Eytan
> Heidingsfeld
> Sent: 29 January 2001 13:14
> To: dyn...@li...
> Subject: RE: [Dynapi-Dev] Next Generation
>
>
> Lets really get the terms straight.
>
> Class-Oriented: You develop classes. You do this be using the prototype of
> an object. Then this object becomes basically a prototype, a
> class and then
> for inheritance an object says use this objects prototype and add these
> members.
> Object-Oriented: You develop classes by defining an interface and then
> implementing this interface. To use this object you create a new object
> based on this class. Inheritance is dealt with by saying I want to inherit
> this and this class. Then your new object has the same members as those
> classes with pointers to their implementation.
>
> There is no PO only CO and OO the difference is inheritance and memory
> management but since there is no memory management in JS there is no
> difference but inheritance.
> 8an
>
>
> _______________________________________________
> Dynapi-Dev mailing list
> Dyn...@li...
> http://lists.sourceforge.net/lists/listinfo/dynapi-dev

Re: [Dynapi-Dev] Next Generation

[Doug M. <do...@cr...>]

An if Netscrape says it it has to be correct?
----- Original Message -----
From: "Cameron Hart" <ca...@bi...>
To: <dyn...@li...>
Sent: Monday, January 29, 2001 6:23 AM
Subject: RE: [Dynapi-Dev] Next Generation


> netscape themselves refer both javascript and java as object-based
> languages. javascript they say is prototype-based, and java is
class-based.
> both are object-based.
>
> straight from the horses mouth
> http://developer.netscape.com/docs/manuals/js/client/jsguide/obj2.htm
>
> > -----Original Message-----
> > From: dyn...@li...
> > [mailto:dyn...@li...]On Behalf Of Eytan
> > Heidingsfeld
> > Sent: 29 January 2001 13:14
> > To: dyn...@li...
> > Subject: RE: [Dynapi-Dev] Next Generation
> >
> >
> > Lets really get the terms straight.
> >
> > Class-Oriented: You develop classes. You do this be using the prototype
of
> > an object. Then this object becomes basically a prototype, a
> > class and then
> > for inheritance an object says use this objects prototype and add these
> > members.
> > Object-Oriented: You develop classes by defining an interface and then
> > implementing this interface. To use this object you create a new object
> > based on this class. Inheritance is dealt with by saying I want to
inherit
> > this and this class. Then your new object has the same members as those
> > classes with pointers to their implementation.
> >
> > There is no PO only CO and OO the difference is inheritance and memory
> > management but since there is no memory management in JS there is no
> > difference but inheritance.
> > 8an
> >
> >
> > _______________________________________________
> > Dynapi-Dev mailing list
> > Dyn...@li...
> > http://lists.sourceforge.net/lists/listinfo/dynapi-dev
>
>
> _______________________________________________
> Dynapi-Dev mailing list
> Dyn...@li...
> http://lists.sourceforge.net/lists/listinfo/dynapi-dev