Re: REXML 2.5.7 and 2.4.7

On 26 Feb 2003 08:18:35 -0800, Sean Russell <[e-mail]> 
wrote:

>> and it is not fair to slam them based on some random version now 
>> everyone
>> has moved on.
>
> Moved on from what?  I didn't say we were using an old version of
> Internet Explorer.  Glaxo-Smith-Klein, one of the largest
> pharmaceutical companies in the world, is still based entirely on
> Windows NT, and won't upgrade to a different version of Windows for
> another two years, at least.  I expect that this is fairly common in
> business communities.  I fail to see why a reasonably recent version
> IE should neccessarily have worse XML and XSL support than, say,
> Mozilla.

Hmm if you are referring to the latest version, fair enough. It is also a 
valid point that many companies are slow to roll out new technology. But 
for a fair comparison you should then compare Mozilla as it were at that 
time. If a company is unwilling to upgrade their browser version, they are 
typically also unwilling to install an entirely different brand of browser.

>> browser versions is a nightmare.
>
> Ideally, you'd be able to write your application based on the
> /standards/ you're using, not how any given third-party piece of
> software interprets them.

Agreed - in this case standards haven been very precise though - and 
vendors haven't made it easier. It's a mess and the situation is far from 
ideal. If you choose following standards you can't create anything near 
what is considered temporary.

BTW: I've just switched to using Opera 7.0 as browser, news and mail client 
- and it is pretty cool so far.

> I distribute REXML, and I try darned hard to make sure that it
> conforms to spec.  And, I do this for free.  My criticism of Microsoft
> was that, with all of their resources *and* all of their revenue, they
> can't seem to keep their software even as conformant as, well,
> anything else.

I perfectly understand from your point of view. I just think you can find 
worse examples than XML and XSLT. If you read about OpenGL you have to be 
impressed with the independence NVidia exercises.
Microsoft need to stay conformant in XML for strategic reasons but need to 
dominate the graphics platform with DirectX.

> I love web services, but I have a real problem with the whole servlet
> paradigm.  People are, essentially, tacking on stateful extensions to
> a stateless protocol, and I find it really distasteful.  On top of
> that, HTML is /not/ a good choice for a UI interface, and until we get
> universal broadband, using HTTP as the communication layer between a
> complex application client and the application logic on a server is a
> terrible idea.  This is way off topic though.

Anyway - if someone want's to write a networked GUI renderer in Ruby with 
focus on OpenGL and GLFX I think I'd be cool and very useful. Add C for 
performance where applicable.
I've some Flash renderers using OpenGL for added performance.

> can reduce network traffic.  On the other hand, there will always be
> some applications that are simply better as thick client applications.
> Email is my favorite example.  Nobody in their right mind who uses
> email a lot would use a webmail gateway as their primary email client,
> if they have an alternative.

But Google News is primarily bad because it is 4 hours delayed - otherwise 
it is the only tool I am aware of the actually sorts the threads in a 
usable way - i.e. thread with most recent post listed first, not sorted by 
original post. (How difficult can it be <sigh>). It's also good at 
displaying long threads.

When I mentioned bad network implementation  - well I've never been 
impressed with Rational products -  I was in particular thinking of 
ClearQuest bug tracker - it's completely useless as a desktop application 
over a 750KB/s WAN link, so we use the Web client over the same WAN link, 
as if the information passsed is any different. The same is true for 
ClearCase source control, except you can have a server at each endpoint. My 
pet favorite is Perforce (source control) which is the ideal networked 
client / server application. OK client GUI could be more flashy, but it 
handles information well.

> Having access to a webmail gateway is
> *nice*, mind you, but even over broadband, is tediously slow compared
> to a thick client.

Yes - mostly because you have a per client database of old mails. There 
isn't really a good reason that web clients should be slow otherwise - look 
at how fast Google servers search pages and news for that matter - but 
granted the have some computing power. Point is the server needs to serve 
the data in whichever way. If we pick Perforce again - it has a local 
client cache of files, otherwise it would be slow as well.
I'd don't care if a client is thick if it's easy to deploy and the client 
data can be easily regenerated - as with Perforce and to an extend email.

> The fact that I have to access usenet through a
> web interface is the the only thing tempting me to sign up for a
> usenet account, so that I can use a real client.
>
> But... for the moment... imagine how nice it would be if you could use
> Ruby as an HTML scripting language.  Mmmmmm.

Nice yes. But I don't think JavaScript is bad. What is bad is, DOM events, 
and the way data flows between frames and pages, and the lack of proper 
session structure that sucks. And Ruby won't change that. However nice Ruby 
may be, it won't make things significantly better.

Therefore, please implement an OpenGL driver client scriptable with Ruby 
and driven by a proper socket connection - using port 80 just for the sake 
of firewalls.
Dump the DOM.

Mikkel