informal /View desktop survey
[1/17] from: jason::cunliffe::verizon::net at: 6-Sep-2002 15:02
I'd love to hear what other rebolers think about View/Desktop.
What's best/worst about it?
How much do you use it?
What would like to improve?
New features?
Is it being surpassed by wikiblogs+xmlrpc/rss/soap ideas?
???
When I first downloaded REBOL/View, I was thrilled especially by the
View/Desktop and the potential for a networked community of rebsites. But as
time went by I found I stopped using it.
I feel it is a great prototype design.. but needs work imo. I do not think it
can succeed until some issues are addressed.
RT are focused on IOS. Perhaps it is a lot of work to improve View, perhaps not.
My impression is that needs a design document describing improvements, how and
why. And marking those which are within easy reach as opposed to the hard ones.
There are likely all sorts of scripts out there now which can be used to improve
Desktop.
Like many aspects of REBOL, there is so much untapped potential already, just
using existing tools and characteristics more intelligently.
What do you think?
What can we do about it?
- search
hard to find things/scripts/sites with it.
google is blind to it
- presentation
one issue is that it only presents rebsites + content as icons. With small
numbers this is fine, but the interface soon becomes a victim of its own
success. To handle larger numbers one needs the option to present content as
lists. Filters for type, metadata, date, source, size etc.. All standard
features in multimedia browsers, but also ones which REBOL could enhance easily.
- history/memory/documentary
hard to see what's new, changed, off-line, deleted, ...
This a big time waster. It should not be hard for Desktop to get status and
provide some graphic feedback upfront about access [colored typestyle or
similar].
- annotate + share
so you find some cool stuff - what's the best way to add metadata, notes
and structure this?
blogs and wikis have shown the value and interest of dynamic web publishing.
What can Desktop learn from them, and vice-versa? How can it be integrated with
a wikiblog system like Vanilla?
./Jason
[2/17] from: brett:codeconscious at: 8-Sep-2002 18:03
Hi Jason,
You've made very good points. I guess a lot revolves around what people
imagine the Desktop is used for (or can be used for). I'd suggest that most
rebsites (including mine) have at least initially seen it as a type of
static browser, but if you add the dynamic/interactive element then it could
gain far more use. E.g I've added a codeconscious.com folder inside of my
local folder - the icons there represent automated aspects of
codeconscious.com website management. So in this case, the desktop becomes a
task based launcher.
In some ways, in these days of flash interfaces (or at least the possibility
of such), it might be considered nice to have more custom visual control
over the desktop by a specific rebsite. E.g if you enter codeconscious.com
you'd pick up the look and feel of codeconscious.com. On the other hand, the
constrained layout we have now allows a user to move around different
rebsites without getting confused on how to operate them.
<Blue sky>
Desktop
in my mind should point to a common working space for disparate
items. It would be interesting to investigate how different rebsites could
offer up resources to the desktop such that the user via the desktop could
combine them in different ways. E.g if rebsites offer "web services" how
could the desktop enable a user to connect different services together?
</blue sky>
As far as status items go the still present reality of dial-up constrains
certain possibilities.
The google issue is interesting. Some of my rebsite content on
codeconscious.com is already the source for generated website HTML. I aim to
make all content on codeconscious.com available via HTML and View/Desktop.
Maybe some sort of generalised rebsite to HTML generator could made.
Interestingly, if a production REBOL browser plugin was available it would
likely work against View/Desktop.
Overall I think the View/Desktop needs somehow to be incorporated into
people's daily routines. Right now you could probably divide up Desktop into
three major territories - RT, Sites, Library. "Library" because it allows
community additions and downloads is probably the closet thing to a service
in Desktop right now. Maybe desktop needs more "services", but which perhaps
are built on top/via community collaboration. This is where your suggestions
of annotate and share come in, I'm just suggestnig an overall name and a
place to plug em in conceptually. So these services things like:
"Knowledge Base" - searchable archive (distributed hopefully).
"Online Expert" - e.g chat to a really keen contributor(s).
"Code Works" - build an app in real time with others.
It would probably be easier to build these on top of IOS, if IOS became more
widely spread. But while that is true, much of the look and feel can be the
same.
My comments are a bit disjointed but I wanted throw in my AUD$0.05 in so
there you have em/it :^)
Regards,
Brett.
[3/17] from: rebol:optushome:au at: 8-Sep-2002 18:36
The desktop is extremely flexible http://www.rebol.com/docs/desktop.html
I've used the View/Desktop for custom delivery. By editing the index and
bookmark files, removing the rebsites list and tools etc, creating custom
backgrounds and all the reblets the client needed for their workflow. So
they have all the scripts they need in a simple convenient space and they
are always uptodate. Their major tasks are listed on the Left (Replacing
REBOL.Com, Local, Console etc) while under each task are the scripts and
documentation required.
Cheers,
Allen K
[4/17] from: jason:cunliffe:verizon at: 8-Sep-2002 16:47
> I've used the View/Desktop for custom delivery. By editing the index and
> bookmark files, removing the rebsites list and tools etc, creating custom
<<quoted lines omitted: 3>>
> REBOL.Com, Local, Console etc) while under each task are the scripts and
> documentation required.
Allen
Thanks. Can you say more about this please?
./Jason
[5/17] from: jason:cunliffe:verizon at: 8-Sep-2002 17:12
Hi Brett
Thanks for your reply..
> <Blue sky>
> "Desktop" in my mind should point to a common working space for disparate
<<quoted lines omitted: 3>>
> could the desktop enable a user to connect different services together?
> </blue sky>
Yes Nice idea.
> As far as status items go the still present reality of dial-up constrains
> certain possibilities.
<<quoted lines omitted: 4>>
> Interestingly, if a production REBOL browser plugin was available it would
> likely work against View/Desktop.
Yes there is ummm contradictory irony to it.
rebsites are hard to search either from google, or from rebol.
That's not right.
I have not yet found a decent rebol technique with google.
For example, how do you best search for a script named "test.r" or "index.r" ?
> Overall I think the View/Desktop needs somehow to be incorporated into
> people's daily routines. Right now you could probably divide up Desktop into
<<quoted lines omitted: 7>>
> "Online Expert" - e.g chat to a really keen contributor(s).
> "Code Works" - build an app in real time with others.
Yes that's good.
I was thinking simple metadata blocks included in rebsites coudl be harnessed in
all sorts of ways, IF Desktop was smarter about looking for them and acting on
them. This is influenced by my ongoing experiments with Vanilla. The structure
is very simple:
content is stored as 'snips' in /space
every .snip file has a matcnhign invisible .metadata file The default Vanilla
uses a small seta of metdata elements for all snips, stored in a block. Thsee
include number oc accesses, the creator and last-editor id etc.. But is is very
easy to add custom fields to the block and write extensions to use them.
Typically these would be invisble 'dyansnips', little rebol scripts stored
moduelarly /apps and exposed to to the rest of the system via a simple syntax:
{brett.list-snips-by-metafield:ideas}
That will call a Vanilla dynascript %apps/brett/list-snips-by-metafield.r and
passit a value "ideas". This example script would search though all block in all
the .metadata files looking for a field called ideas:
Then it would reurn linked HTML list for display of the snips matching those.
This is very easy to do. Thanks to REBOL set and get is trivial and very fast.
I can imagine applying the same technique to Desktop and rebsites.
AS you say so that "somehow to be incorporated into people's daily routines"
> It would probably be easier to build these on top of IOS, if IOS became more
> widely spread. But while that is true, much of the look and feel can be the
> same.
I have still not gotten to grips with IOS.
So manybe I am reinventing some wheels here//
> My comments are a bit disjointed but I wanted throw in my AUD$0.05 in so
> there you have em/it :^)
No thank you. All comments are welcome.
./Jason
[6/17] from: kemp:extelligence at: 8-Sep-2002 20:08
Out. co. is building metadata servers precisely for information-finding.
We are supporting standards such as Dublin Core, Microsoft Properties, Adobe
XAP/XMP, and many more, including Semantic Web triples. I'd be quite willing
(and eager) to add a Rebol-specific metadata format to the product if any of
you would like to work together to define a 'standard' for Rebol metadata.
This could eventually be exposed to both Web and Rebol through standard
search engines, and RSS, Z39.50, etc.
Kemp
[7/17] from: chris:ross-gill at: 8-Sep-2002 20:21
Hi Kemp,
> you would like to work together to define a 'standard' for Rebol metadata.
I thought the REBOL Header was standard meta-data...
- Chris
[8/17] from: kemp:extelligence at: 8-Sep-2002 21:22
Certainly that's a starting point; however, the Rebol header block is
extensible, not mandatory, and not really typed in any way. My first thought
is a func in a well-known .r file that mandates certain metadata that would
be understood by the engine, and that would send it to the engine and
publish it.
K.
[9/17] from: carl:cybercraft at: 9-Sep-2002 15:54
On 09-Sep-02, Jason Cunliffe wrote:
> I have not yet found a decent rebol technique with google. For
> example, how do you best search for a script named "test.r" or
> "index.r" ?
I've just looked for a file-extension search engine and found this...
http://extsearch.com/
but it knows of no .r files. Hmmm, and it's not a web searcher
anyway, just a searcher of their database of extensions. (I
submitted .r to them anyway.)
Odd. I can't find any attempt at a file-extension searcher anywhere.
AltaVista though, if you search for "rebol" will offer some
suggestions to refine your search, the suggestions currently being:
Carl Sassenrath, Dummies, Ingo, Official Guide, Perl, Platforms,
Ralph, Rebol Scripts, Rebol Technologies, Rolls, Scripting Language,
and User's Guide.
(Unfortunatly, searching for "perl" doesn't offer you "rebol" as a way
to refine your search:)
It's "Rebol Scripts" are just html pages though, not .r files. Which
is fair enough I guess, since it's an engine for web-browsers.
My solution to the problem would be to add a menu item to the View
Desktop that launches Frank Sievertsen's reb-search script. (Anyone
can do this - extending the menus is part of Desktop's design.) It's
pretty primitive in the interface department, but it does search the
REB quite well. This doesn't solve the problem of how to find
scripts not linked to by the Desktop, but it's a start.
I think it's up to us to write search engines for REBOL scripts, and
if we want them findable by reb-browsers, have web-pages linked to
those search engines, or at least make a nicely organised directory
to the REBOL scripts they find.
--
Carl Read
[10/17] from: jason:cunliffe:verizon at: 8-Sep-2002 22:15
> Certainly that's a starting point; however, the Rebol header block is
> extensible, not mandatory, and not really typed in any way. My first thought
> is a func in a well-known .r file that mandates certain metadata that would
> be understood by the engine, and that would send it to the engine and
> publish it.
>
> K.
Hi Kemp,
Do you mean something like :
REBOL [
File: %index.r
Description: "Rebsite index with MetaData in block header"
Author: [k--thecastle--prague--org]
Date: 2002-09-08
Metadata: [--keywords and sub-blocks here--]
Engine: %rebsites.net/engine.r
]
...Where the engine.r would be a sort of rebolistic version if XML-DTD but a lot
nicer :-)
And the Metadata: block in the header would be picked up by any interested
service.
The remote service would have the choice to use its own engine and/or load or
compare against the one defined in the script header block. Something to allow
easy convergence or divergence as application or access demands.
Then at a minimum engine.r might check for a basic metadata block. A small
script or View/app could be run to invoke the engine check that files and help
people to quickly accurately add/edit their metadata fields without having to
remember detailed syntax or get too low down..
The engine.r loaded in the desktop app might also check for obvious conflicts
between a local engine and a remote one... [which could lead us to the hairy
topic of topic maps]
The key point being a reasonable separation of content and tool, but exploiting
REBOL lovely way of fusing them ;-)
hope I am making sense here..
./Jason
[11/17] from: carl:cybercraft at: 9-Sep-2002 16:14
On 09-Sep-02, Kemp Watson wrote:
> Certainly that's a starting point; however, the Rebol header block
> is extensible, not mandatory, and not really typed in any way. My
> first thought is a func in a well-known .r file that mandates
> certain metadata that would be understood by the engine, and that
> would send it to the engine and publish it.
> K.
keywords for starters, perhaps?
REBOL [
keywords: ["this" "that" "the other"]
]
One advantage of keywords is it doesn't need a lot of thought to think
up a few words related to your script. This means people are more
likely to bother to add them. Certainly easier than writing a
description of your script...
>> Hi Kemp,
>>> you would like to work together to define a 'standard' for Rebol
>>> metadata.
>> I thought the REBOL Header was standard meta-data...
>> - Chris
--
Carl Read
[12/17] from: greggirwin:mindspring at: 9-Sep-2002 0:05
Hi Kemp,
<< Out. co. is building metadata servers precisely for information-finding.
We are supporting standards such as Dublin Core, Microsoft Properties, Adobe
XAP/XMP, and many more, including Semantic Web triples. I'd be quite willing
(and eager) to add a Rebol-specific metadata format to the product if any of
you would like to work together to define a 'standard' for Rebol metadata.
This could eventually be exposed to both Web and Rebol through standard
search engines, and RSS, Z39.50, etc. >>
That sounds great to me! You've gotten some informal responses already but I
think this would be a great opportunity for us (the REBOL community) to come
up with something that can be used not only in the context of the View
desktop, but for much much more.
Sadly, my free time is in short supply lately and I know very little about
the meta-data initiatives you mentioned (looked at Dublin Core stuff briefly
a while back). I hate to ask more of you when I can't offer much myself, but
it seems like you would be a great champion for this project.
If you could provide a few important links, and maybe a very quick review of
what you see as the pros and cons for each of the standards you're
supporting already, I think the folks here could come up with some really
good input for you. Your expertise would be valuable in separating the wheat
from the chaff and finalizing the spec.
So, without asking you for any kind of commitment, I think this could be a
very worthwhile project and I hope something comes of it!
--Gregg
[13/17] from: kemp:extelligence at: 9-Sep-2002 2:24
Yes, that's very much along the lines of my thinking, as is Carl's mention
of keywords.
A good place to start would perhaps be
http://www.dublincore.org and
http://www.dublincore.org/documents/1999/07/02/dces/
Another crack at it:
REBOL [
File: %index.r
Description: "Rebsite index with MetaData in block header"
Author: [k--thecastle--prague--org]
Date: 2002-09-08
Metadata: [
%rebsites.net/dublincoreengine.r [
title "some title"
creator "K. Watson"
subject "Rebsite, indexing"
description "Rebsite index with MetaData in block header"
publisher "K. Watson & Friends Inc."
contributor "J. Smith"
contributor "Hay Uthere"
date/created 2002-09-04
date/modified 2002-09-08
type "Software"
identifier http://wherethislives.com
language "en-US"
rights "K. Watson & Friends Inc."
]
%anotherrebsite.org/generickeyvalueengine.r [
'somekey 5
'someotherkey "a text value"
'anotherkey [ block values ]
'andonemore 'key
]
]
]
[14/17] from: kemp:extelligence at: 12-Sep-2002 2:29
XFML looks to be quite interesting - based on RDF, which has the potential
to tie it to the Semantic Web. There's been some criticism however that 1)
RDF is less rich than RSS, and that the Semantic Web is doomed to failute if
it uses RDF, and 2) that topic maps don't use the full potential of RDF.
I dunno, gotta read more.
I WILL play with this stuff in REBOL.
K.
[15/17] from: jason:cunliffe:verizon at: 12-Sep-2002 4:41
The Weblog MetaData Initiative
http://www.truthlaidbear.com/blogmd/ConceptualDM.html
[16/17] from: jason:cunliffe:verizon at: 12-Sep-2002 1:53
Hi Kemp, Sorry about the delay in responding. Yes I like your version...
I'll dive into dublin core. Meanwhile, have your heard about Peter Van Dijck's
XFML?
I'd like to get you opinion since you know about XFML, and how well you think it
relates to REBOL metadata engineering
XFML: [EXchangeable Facet Map Language ]
http://xfml.org/
XFML allows for easy creation of advanced, automatically generated navigation
for your website. You can even automatically generate links to related topics on
other websites. It also allows for merging of metadata between different
websites.
XFML is designed to be easy to understand, and easy to code for, yet powerful
and flexible. XFML is an open, free format, like SOAP or RSS, and is based on
the topicmaps standard.
The tutorials are handy..
http://xfml.net/index.php?page=XfmlTutorials
./Jason
[17/17] from: jason:cunliffe:verizon at: 12-Sep-2002 4:07
> XFML looks to be quite interesting - based on RDF, which has the potential
> to tie it to the Semantic Web. There's been some criticism however that 1)
> RDF is less rich than RSS, and that the Semantic Web is doomed to failute if
> it uses RDF, and 2) that topic maps don't use the full potential of RDF.
I don't quite follow that last sentence. Probably bacause my grasp is still slim
on the most of these XML meta-kits. Can you elaborate please
Here's some brief comments about XfmlAndOtherTechnologies
http://xfml.net/index.php?page=XfmlAndOtherTechnologies
> I dunno, gotta read more.
me too..
Peter Van Dijk has set up a mailing list for XFML. A good place to put any of
your questions, since there is small but appreciative audience who are really in
to this stuff. His site and blog are worth visiting:
http://poorbuthappy.com/ease/ [yesteday's account of his MS interview]
http://petervandijck.net/
> I WILL play with this stuff in REBOL.
>
> K.
Great to hear that:-)
My impression is that XFML might be very well suited to a REBOL implementation
and vice versa
./Jason
Notes
- Quoted lines have been omitted from some messages.
View the message alone to see the lines that have been omitted