[ng-dhtml] markup languages
ng-dhtml at dept-z.com
Tue Sep 21 13:52:51 CDT 2004
Hi all, I'm back.
One addendum to what Mark had to say (so we're clear)...XAML is less a new
markup language and more a markup version of the actual .NET Avalon/Win
forms model. Essentially it just takes what you'd do within code, and get a
markup version from it. From a technical standpoint, it's essentially a
SOAP-formatted serialization of what you might do with straight-up code.
Because of that, it *can* do almost anything you'd do with code itself; the
concept is that you have a markup language that can be deserialized and
compiled on the fly.
I would say that while the approach is pretty straight-forward, it is not
something that we should try to code against at all.
But, on the other hand, there's something to be said for the approach of
developing a markup model that is simply a serialized version of the object
model behind it. I *have* done similar things in JS, and it can be a really
powerful approach; the one caveat is that the markup would end up being
proprietary. Not something that I object to personally, but I can see
problems with it.
As for the others, I haven't looked too close at the Lazlo systems deal; I
have looked at Macromedia's, and found it lacking (in the same way that
their XML model is a bit lacking); and I am in total agreement with Mark
about the cluster-f*ck that is the Moz XUL implementation.
I'm wondering if coming up with a markup language first--then making it
work--is the wrong approach...it may be better to come up with the
underlying object structure first--while keeping an eye on the serialized
version--so that things are essentially seemless, and more importantly,
extensible without rewriting a good deal of code.
> -----Original Message-----
> From: NG-DHTML-bounces at netwindows.org
> [mailto:NG-DHTML-bounces at netwindows.org] On Behalf Of Mark D. Anderson
> Sent: Sunday, September 19, 2004 7:50 PM
> To: discussion on the future of DHTML
> Subject: Re: [ng-dhtml] markup languages
> i don't think any of the extant xml-based gui declaration
> languages are satisfactory enough to be relied upon for our "native"
> i still think it is useful to mine the others for ideas, and
> be as close as possible.
> They almost all have crippling platform issues, which is why
> we are talking about Dojo at all (they require a flash
> runtime, or a java runtime, or a XUL runtime, etc.).
> But their representations aren't bad.
> They all have something to offer.
> XUL has probably the most powerful support for separating the
> selection of data from how it should be shown.
> On the other hand, XUL has basically no data binding.
> (Actually, if you dig deep enough you will find this
> RDF-based horror that I can't even describe without the
> nausea overcoming me.)
> Of XAML/XUL/Laszlo/Flex, only Laszlo uses xpath in a
> reasonable way in data binding. Even then, they only use a
> sort of "almost xpath" syntax with variable names prefixing
> They all differ considerably in how they address binding of
> gui event handlers, styling, remote data services, inclusion
> of client-side css or script code, etc.
> btw, some interesting smorgasbord pages are:
> (Greg Bauer has this bizarre fixation on calling everything XUL;
> just ignore that)
> I'd suggest that after a due amount of study (which you may
> have already completed), just start writing some sample
> components in a vocabulary you make up, to implement some
> demo app you already have in mind. Then that can be molded
> back to similarity to existing precedents as need be. Sort of
> like building the chunnel.
> I'd be interested in roundtrip translation with any xml gui
> vocabulary that has an open source visual gui builder that
> (a) is platform agnostic, and (b) doesn't suck.
> So far, as far as I can tell, there aren't any.
> There are a whole bunch of eclipse plugins (VE, V4ALL,
> Eclipse Designer, WebOnSwing, Big Splash, etc.) but so far
> they all seem to be a joke relative to what commercial gui
> builders can do.
> Since IBM seems to be behind VE, maybe that'll be the horse
> to ride, to ultimately get something akin to Flex Builder.
> (Glade doesn't count because it fails (a); it isn't platform
> agnostic and doesn't have a clean independent gui markup language.)
> NG-DHTML mailing list
> NG-DHTML at netwindows.org
More information about the NG-DHTML