eggfree at eggfree.net
Mon Sep 24 14:48:16 PDT 2007
Not that I have a vote, but I'd put it around 0--
At the common risk of making a fool of my self here goes...
there are positives and negatives to both, although the first question
in my mind would be about the extent and use of inheritance. There are
good time to use it for sure, but if sanctioning this approach implies
"overusing" inheritance that may prove inappropriate too. Some of the
contributors Matthew may speak of object more to the imposition of a
java model on a language that if anything is more perl like.
Before instituting too much superclassing it may make sense to adopt
whatever mechanism DOJO uses to do documentation -- one of the things
that makes java workable is the high level of doc integration into IDE's
-- to discover what methods are available and how they apply.
(I say that as more a java person)
Just my 0.02
Matthew Eernisse wrote:
> -1 on this.
> I object to this for a couple of reasons.
> Technical grounds:
> 1. Perf -- this adds more indirection and code execution. Our UI code
> is already slow, and we need to be making it faster, not slower.
> 2. Stability -- we picked Dojo to do some very specific things (data
> transport, events, pub/sub, packaging, build), and are already going
> to have to do a bunch of rewriting because they rewrote the toolkit
> and broke compatibility.
> I would be really reluctant to build in even more dependence on Dojo,
> especially for something as ubiquitous as how we define our business
> Social grounds:
> There is a class of developer who is not particularly familiar with
> Dojo, or interested in doing everything in their code with the Dojo
> API. I am a fan of Dojo, I think expecting something that low-level to
> be done the Dojo way will be very offputting to those people -- people
> we'd like to recruit to help us hack the Cosmo UI.
> I also count myself in that group -- unless there's some specific,
> technical benefit, I myself strongly prefer standard, idiomatic
> Travis Vachon wrote:
>>> Dojo provides us with a nice solution to the manual labors of
>>> inheritance, I think we should standardize on it.
>>> cosmo-dev mailing list
>>> cosmo-dev at lists.osafoundation.org
>> cosmo-dev mailing list
>> cosmo-dev at lists.osafoundation.org
> cosmo-dev mailing list
> cosmo-dev at lists.osafoundation.org
More information about the cosmo-dev