Re: [Dynapi-Dev] DynAPI3 beta release - few comments

[Dan W. <da...@wi...>]

On Tue, 2003-04-29 at 09:52, Joy Ride wrote:
> Hello,
> 
> I would like to make few notes about comming release of DynAPI3. I'm sorry 
> for that I haven't gone trough DynAPI3 code so that I would know wheter 
> these are allready taken count of. These are based on our experience and if 
> any one of you have better knowledge if these issues, I would be happy to 
> know about them.
> 
> 1) There should be attention paid in to image size/sizing issues in DynAPI3. 
> (At least in DynAPI2 there were several problems related in it.) Netscape 
> 6/7 and Opera browsers are able to find out image dimensions only after the 
> whole document is loaded. There for DynAPI should employ mechanism which 
> prevents image size queries before document is loaded on these browsers. 
> Exspecially DynAPI core shouldn't rely on image dimensions before document 
> is fully loaded.
> 
> 2) The Javascript engine of Netscape is very slow on what comes to events. 
> Special atention should be paid in to performance issues in Netscape 
> browsers (more shortcuts / performance tweaks?). Exspecially dragging is a 
> problem in Netscape. Jukka suggested that it might be good idea to consider 
> different approaches to click events and drag events in Netscape (simplified 
> drag module for Netscape?).

I don't know if this is possible, but maybe it would work better if we
use some more native netscape 6+ objects, instead of the standard dom. 
Netscape has 'behaviors' in their css files for the browsers themselves
to call javascript function, say on mouseover of an element, etc.

> 
> 3) In Netscape 4 loading is the hardest issue. Fairly badly documented 
> feature describes that Netscape 4 can have only one layer open for loading 
> at the time. Other layers have to be closed before loading stuff in to other 
> layer (note that Netscape doesn't close layer automaticly). If something is 
> tried to load in to layer when other layer is open Netscape crashes. This 
> suggests that some sort of loader is needed (based on our testes, in some 
> browsers onLoad detection is not 100% acurate).
> 
> 4) Opera 5 is very slooooooow... it may affect in to some script 
> functionality.
> 
> 5) Definition of looks should be made with styles when ever possible. Style 
> definitions should be done based on DOM standard.

I totally agree with this, unless the current standards(CSS&DOM) are
very bad in depicting a certain value and what it means, I say stick
with the standards.

We could also look at future standards values, like what is defined in
CSS3.

>  Which means that widht 
> equals in to width of content area. Borders are excluded from this width 
> (this behavior differes from DynAPI2 approach). Following browsers use DOM 
> standard in a way described earlier (others need to be fixed "manually" in 
> dynapi): Opera from version 6 on, Netscape from version 6 on, MSIE from 
> version 5 on machintosh platfom and MSIE from version 6 on PC platform. On 
> MSIE6 on PC platfor ther is one catch though. MSIE6 uses DOM behavior only 
> if there is correct doctype definition (=DOM) and link to correct standard 
> in HTML file header. Otherwice it uses old aproach, where borders (for 
> instance) are counted as part of widht. There for there should be (?) some 
> sort of detection for doctype definition in DynAPI.
> 
> These are the issues, we think that should be dealt with DynAPI3 before 
> betarelease if they aren't dealt allready. I think that we will be able to 
> provide the results of browser compability tests (Excel files, we have 
> tested all the objects and methods separately on each browser) shortly... I 
> will let you know when we are ready with those.
> 
> 
> - Juho Risku / Helmi Staff
>   http://www.visualway.com/helmi
> 
> ___________________________________

Just a few thoughts....
-- 
Dan Willemsen <da...@wi...>