From: Barre B. <ba...@ho...> - 2000-12-01 10:16:18
|
> I have completed splitting the API into parts for IE and NS, even to their > specific versions. I have nodiced a 10x increase in speed. And a great > increase in stability. I believe that the current API model is a very sound and applicable one. It is coherent and easily followed. But.... Speed and stability easily outweigh ease of implementation. This I have concluded trying to create complicated webapps in dHtml. Another thing is that a split would decrease the amount of browser checking in the API, which would also speed things up. > If the community was to appoint a maintainer for each browser, someone who > is highly sufficient in the capabilities of the browser, then they can > optimize their versions without fear of gumming up the works for other > browsers and platforms. This is a VERY valid point. Trying to keep platform compatability intact while developing is a royal pain in the ass. > I'm not staying it will be easy. But I believe that the benifits would > outway the work it would entail. This I totally agree with. / Bart > ----- Original Message ----- > From: "Abel Eduardo Cantu Salas" <abe...@al...> > To: <dyn...@li...> > Sent: Thursday, November 30, 2000 2:00 PM > Subject: RE: [Dynapi-Dev] inheritance crusade / SuperClass stuff > > > > > But isn't it better to use optimised versions for the browser, not > > > the one for all? > > > > It sound very interesting indeed Alexey... then you can also create new js > > for newer browsers whithout any need to touch other js (just like dlls or > > device drivers, you'll load what you need); it will allow you to reduce > the > > amount of code that the client most download and most important than that > is > > the fact that this code will be really very optimised. However this has an > > important drawback that we most consider: > > > > Doing this means an extra work in trying to maintain standardized all the > > APIS for diferent platforms and diferent browsers, id est, if we add > > something to an API we must add the same functionality to all other; and > not > > just that, separating everything so you can then optimize a lot, could > make > > easy a loss of control in what is disponible for what platform. > > > > Im sure we don't want a DynAPI that says: "this is disponible only for NS" > > or vice versa, then we most work a lot to keep things in the right path. > > > > Any opinions or sugestions about this matter?, there is a lot of > advantages > > and drawbacks in a schema like this... > > > > -----Original Message----- > > From: dyn...@li... > > [mailto:dyn...@li...]On Behalf Of Alexey > > Medvedev > > Sent: Jueves, 30 de Noviembre de 2000 10:42 a.m. > > To: dyn...@li... > > Subject: Re: Re: Re: [Dynapi-Dev] inheritance crusade / SuperClass stuff > > > > > > On Thu, 30 Nov 2000, Barre Bizon wrote: > > > > Yes - I told it is JS_1.3_ > > > > > Another thing... neither the function.call() nor the > > > function.apply() method work under IE4+. > > > Unfortunately. > > > > Also add .watch() to the list :-/ > > > > But isn't it better to use optimised versions for the brouser, not > > the one for all? > > > > So main "dynapi" whould be a selector of "dynlayer_ns4.js" > > or "dynlayer_ie4.js" but keeping same API? > > > > I thinks using "new Layer" in NN4 is better then creating > > lots of CSS and <div>-s (memory usage and resistance to Resize): > > > > http://deep.kiev.ua/JS/WebGram/ - resize resistant without > > any reloads. > > > > ps. last one i need to make site as one page , with changed layers > > Same as www.htmlguru.com is. > > > > Malx > > > > _______________________________________________ > > Dynapi-Dev mailing list > > Dyn...@li... > > http://lists.sourceforge.net/mailman/listinfo/dynapi-dev > > > > _______________________________________________ > > Dynapi-Dev mailing list > > Dyn...@li... > > http://lists.sourceforge.net/mailman/listinfo/dynapi-dev > > _______________________________________________ > Dynapi-Dev mailing list > Dyn...@li... > http://lists.sourceforge.net/mailman/listinfo/dynapi-dev |
From: Cameron H. <ca...@bi...> - 2000-12-05 10:51:50
|
I posted a suggestion to Dan's message about using a Makefile system for building a final version of the code, I thought it would be worth suggesting it to this thread of the discussion. I think it would be a good idea to have a C style #define in the src javascript to allow building final versions for different browsers. I think the advantage of doing this is the cross-browser API remains in one set of source, and users people can then build either seperate versions for each browser, or build them all into one library. From a development point of view, to me it makes more sense to keep all the code in a canonical form, rather than split for every major browser/platform that comes into existence. Cam. > -----Original Message----- > From: dyn...@li... > [mailto:dyn...@li...]On Behalf Of Eytan > Heidingsfeld > Sent: 05 December 2000 09:24 > To: dyn...@li... > Subject: Re: [Dynapi-Dev] Splitting the API? > > > Although I am not so sure why thing would run so much better with the API > split I also have 2c. Basically everything that is browser based are the > actual low-level components (DynLayer, events....) why don't we > split those > components in to one for each version. Then when a higher level > component (a > decendent) is created in the constructor you add the following code > if (browser == IE){inherit from IEDynLayer;} > if (browser == NS){inherit from NSDynLayer;} > .. > .. > .. > You get the picture. Then although it might be a hassle to orginize the > low-level comps the higher level one will be relitivly easy. > > 8an > > _______________________________________________ > Dynapi-Dev mailing list > Dyn...@li... > http://lists.sourceforge.net/mailman/listinfo/dynapi-dev |
From: Robert R. <rra...@ya...> - 2000-12-07 00:25:32
|
I think right now there a bigger issues. Such as bugs in the pushpanel, scrollpanel, and loadpanel. If we can make this version stable, then creating a parser to separate the code would be nice. But, it really is not the most important problem right now. We need to release a new version soon, but we can't until it becomes stable. -- // Robert Rainwater On 12/5/2000, 5:51:29 AM EST, Cameron wrote about "[Dynapi-Dev] Splitting the API?": > I posted a suggestion to Dan's message about using a Makefile system for > building a final version of the code, I thought it would be worth suggesting > it to this thread of the discussion. > I think it would be a good idea to have a C style #define in the src > javascript to allow building final versions for different browsers. I think > the advantage of doing this is the cross-browser API remains in one set of > source, and users people can then build either seperate versions for each > browser, or build them all into one library. From a development point of > view, to me it makes more sense to keep all the code in a canonical form, > rather than split for every major browser/platform that comes into > existence. > Cam. >> -----Original Message----- >> From: dyn...@li... >> [mailto:dyn...@li...]On Behalf Of Eytan >> Heidingsfeld >> Sent: 05 December 2000 09:24 >> To: dyn...@li... >> Subject: Re: [Dynapi-Dev] Splitting the API? >> >> >> Although I am not so sure why thing would run so much better with the API >> split I also have 2c. Basically everything that is browser based are the >> actual low-level components (DynLayer, events....) why don't we >> split those >> components in to one for each version. Then when a higher level >> component (a >> decendent) is created in the constructor you add the following code >> if (browser == IE){inherit from IEDynLayer;} >> if (browser == NS){inherit from NSDynLayer;} >> .. >> .. >> .. >> You get the picture. Then although it might be a hassle to orginize the >> low-level comps the higher level one will be relitivly easy. >> >> 8an >> >> _______________________________________________ >> Dynapi-Dev mailing list >> Dyn...@li... >> http://lists.sourceforge.net/mailman/listinfo/dynapi-dev > _______________________________________________ > Dynapi-Dev mailing list > Dyn...@li... > http://lists.sourceforge.net/mailman/listinfo/dynapi-dev |
From: Brandon M. <bnd...@ho...> - 2000-12-07 03:18:39
|
ACK! That would make development a nightmare.. Change the original code... run it through the code splitter, run the test... I like just changing the code, hit refresh on browser page... much quicker... one less window. ----- Original Message ----- From: "Cameron Hart" <ca...@bi...> To: <dyn...@li...> Sent: Tuesday, December 05, 2000 5:51 AM Subject: RE: [Dynapi-Dev] Splitting the API? > I posted a suggestion to Dan's message about using a Makefile system for > building a final version of the code, I thought it would be worth suggesting > it to this thread of the discussion. > > I think it would be a good idea to have a C style #define in the src > javascript to allow building final versions for different browsers. I think > the advantage of doing this is the cross-browser API remains in one set of > source, and users people can then build either seperate versions for each > browser, or build them all into one library. From a development point of > view, to me it makes more sense to keep all the code in a canonical form, > rather than split for every major browser/platform that comes into > existence. > > Cam. > > > -----Original Message----- > > From: dyn...@li... > > [mailto:dyn...@li...]On Behalf Of Eytan > > Heidingsfeld > > Sent: 05 December 2000 09:24 > > To: dyn...@li... > > Subject: Re: [Dynapi-Dev] Splitting the API? > > > > > > Although I am not so sure why thing would run so much better with the API > > split I also have 2c. Basically everything that is browser based are the > > actual low-level components (DynLayer, events....) why don't we > > split those > > components in to one for each version. Then when a higher level > > component (a > > decendent) is created in the constructor you add the following code > > if (browser == IE){inherit from IEDynLayer;} > > if (browser == NS){inherit from NSDynLayer;} > > .. > > .. > > .. > > You get the picture. Then although it might be a hassle to orginize the > > low-level comps the higher level one will be relitivly easy. > > > > 8an > > > > _______________________________________________ > > Dynapi-Dev mailing list > > Dyn...@li... > > http://lists.sourceforge.net/mailman/listinfo/dynapi-dev > > _______________________________________________ > Dynapi-Dev mailing list > Dyn...@li... > http://lists.sourceforge.net/mailman/listinfo/dynapi-dev |