At 11:54 AM 5/27/2000 -0700, you wrote:
>Hi folks:
>
>With that said, I would be interested to learn more about the existing GUI
>client. What are the group's specifications for the UI in terms of core
>functionality, features, extensibility? Who is currently in charge of
>defining/implementing it? The GUI class mentioned, is it just a quick hack
>to insert and retrieve files? (is it really just one class?) How about
>locating files on the network? I sense that you guys don't want to commit to
>Swing - have you considered maybe a Web interface? Questions, questions...
>

Actually, I'll admit that I've hung around a couples weeks and offered
suggestions "here" without first asking these questions - I have no idea if
anyone is in charge of developing the GUI. If I've been rude, I apologized.
Still, I'll repeat the ideas I've submitted - now connected to a real,
existing application. 

I don't know if anyone has look at the UI I recently submitted but I would
like to think it is a bit more than a quick hack. While currently it mostly
appears as merely a relatively web-bbs view, it is altogether a
sophisticated and extensible interface. The client has it's own
interactivity language that can be used to specify how a series of
documents are put together into a view. And it lets see stuff other folks
posts to freenet automatically. 

Now since freenet is a text-retrieval engine, making it "look like the web"
is fairly trivia - all you need is a web server run at the client machine
with perhaps a little filtering. (but we need to agree on what the exact
protocol of a freenet URL would be - sure "freenet://Key" but what's the
terminator of the key?) Indeed, a local web server makes more sense to me
than a mozilla plug-gin since the user wouldn't have to install a large
application. Imagine if freenet eventually had a "one floppy" set-up size.
"Any web brousser" seems like the ideal target for most of the document
that will be one free - these will be broussed, played or saved and all the
existing apps will do that just fine. 

The trick is to allow enough interactivity so that when someone posts a
document or composes a "freenet page", they can also post a link to that
page on freenet itself and have people see the existence of the new page.
The client I mentioned allows someone to do that. And it is working now
(though most of my testing has been simulation - I have posted and viewed). 

This existing client can be extended to achieve most of the other aspects
of the classic CGI applications (a hits counter would be a bit wasteful)and
beyond to the classic "Xanadu" type features - two way links, all pages
annotatable and so-forth. Basically, if a semi-programmable application is
the viewer, the dynamism of most interactive web pages can be added by
simply adding a program as your basic document. Updatability of documents
can be done this way as well (keyVer1, keyVer2...or KeyTimeStamp345543step345)

Now, the one reason that I'd like folks to notice this is that the system
will only work if a single interactivity language/client system becomes
standard. 

Of course, single freenet documents are either static or expire in a fairly
uncontrollable fashion. The views or "compound documents" that any client
would use to provide updates on new documents would be composed using
agreed ranges of values such as Key1, Key2 etc. (Though my client uses a
similar but more complex system which allows an extensible tree structure.
Keep in mind that this is a system I have worked on - full time - for about
six months - though most of the time not knowing anything about freenet. I
think of it as having fairly rich functionality). 

So, if folks are going to be creating "compound" documents by adding
documents having related keys (key1, key2...keyN+1), there would have to be
a universal "protocol,"  a language, which describes how these things are
found and viewed. Considering we HAVE one such language, I would suggest
that it would be good to standardize on it because without such a language,
people wouldn't be able to create "compound documents" and thus all link
updates to freenet would have to come from the outside - which is a
definite VULNERABILITY and we don't like vulnerabilities. 

The only OTHER thing besides making an existing language standard that
would make sense to me would be to wait until something more standard like
XSLT can accomplish similar things. But this may be a while. And the point
is to have a single updatable viewer. Then the language can be updated over
time - look at the evolution of HTML.

OK, I released the code yesterday, so maybe I'm being touchy. But I really
hope to get some feedback about it. 
(PS, it's at /www.crucialquestions.com/ccq_fnv.zip and it's open source, of
course, submit for your consideration) 

Joe




























_______________________________________________
Freenet-dev mailing list
Freenet-dev at lists.sourceforge.net
http://lists.sourceforge.net/mailman/listinfo/freenet-dev

Reply via email to