[dojo-contributors] Prototype-al simplicity

Bill Keese bill at dojotoolkit.org
Thu Apr 6 23:09:25 EDT 2006

Paul - I agree completely w/what you wrote, and thanks for doing that 
research, especially the part about our version of prototype somehow 
ending up 2-4 times bigger.

How come it's so much bigger?  My guess is that it's NOT because "too 
many modules are being pulled in", but rather something more 
fundamental.  Not sure what though.


Paul Sowden wrote:
> On 6 Apr 2006, at 02:15, Bill Keese wrote:
>> I'm hesitant to start on any new projects when 0.3 is already months 
>> behind schedule.  Having said that...
> Well 0.3 should be pushed.  If it gets to the point were we can't move 
> forwards the release schedule is unrealistic and flawed.
> Introducing these changes is something that I'm incredibly passionate 
> about and I'm psyched that there's momentum to get there.  In my eyes, 
> making Dojo distributions like this makes us relevant, if not to the 
> whole Web then at least to me.
>> I would make "dojo-lite" (a.k.a core) bigger.  Maybe this:
>> bootstraps
>> dojo.lang
>> dojo.event
>> dojo.io  (not sure about the multiple IO packages; seems like one 
>> should be sufficient)
>> dojo.json
>> dojo.dom
>> dojo.html
>> dojo.style
>> I imagine that's similar to what's in prototype but can someone else 
>> comment?
> I think this is as similar to Prototype as we can get.  Prototype is a 
> pretty horrible library for shitting all over the environment.  I think 
> we have enough in these libraries to have a killer core distribution.
> The main stinging point is that Prototype is 48kB *uncompressed*.  
> Building for us, we're 196kB uncompressed and it only shrinks to 116kB 
> compressed.  We're still well over twice the size, and that's not like 
> for like.
> The build file we're looking at is:
>   var dependencies = [
>     "dojo.io",
>     "dojo.io.BrowserIO",
>     "dojo.io.cookie",
>     "dojo.event",
>     "dojo.json",
>     "dojo.dom",
>     "dojo.style",
>     "dojo.html"
>   ];
> Which causes these files to be included:
>   ../src/bootstrap1.js
>   ../src/bootstrap2.js
>   ../src/hostenv_browser.js
>   ../src/string.js
>   ../src/string/common.js
>   ../src/string/extras.js
>   ../src/lang/common.js
>   ../src/lang/type.js
>   ../src/lang/extras.js
>   ../src/lang/array.js
>   ../src/lang/func.js
>   ../src/lang/repr.js
>   ../src/io.js
>   ../src/io/BrowserIO.js
>   ../src/io/cookie.js
>   ../src/uri/Uri.js
>   ../src/json.js
>   ../src/undo/browser.js
>   ../src/event.js
>   ../src/AdapterRegistry.js
>   ../src/graphics/color.js
>   ../src/dom.js
>   ../src/style.js
>   ../src/html.js
> Looking at this it seemed an utterly fruitless exercise fragmenting lang 
> and string to the goal of including less and so overall a detrimental 
> change.  I don't even know what an AdapterRegistry is... anyhow.
> Overall I think this is a solid and noble set of modules to include as a 
> dojo-lite and should make us very attractive, I know it will to me.
> Basically I'd like opinions on this, is this what people were gunning 
> for when we talked about the issue or am I on my own path?  I'd very 
> much like to decide on this so we can move onto the next stage of 
> getting us there with these modules.
> Cheers.
> --Paul Sowden
> _______________________________________________
> dojo-contributors mailing list
> dojo-contributors at dojotoolkit.org
> http://dojotoolkit.org/mailman/listinfo/dojo-contributors

More information about the dojo-contributors mailing list