Mailing List Archive: 49091 messages
  • Home
  • Script library
  • AltME Archive
  • Mailing list
  • Articles Index
  • Site search
 

Some ideas to collaborative environments

 [1/7] from: jsc::dataheaven::de at: 23-Sep-2000 0:45


We should not invent thousands of new protocolls for a collaborative environment. Why not using FTP,HTTP,MAIL, IRC and so on in a highly integrated manner. 1) The notion of "Personal Space" and "Group Space": I do not want to know where my globally accessible files really are stored. I want to sit down on a computer somewhere in the world and have immidiate access to "my" files. My globally accessible files should be saved in a secure manner. I call the virtual space where all my personal files are globally accessible my "Personal Space". The same rules of the Personal Space counts also for the Group Space . A group are some people that share some files and have need for efficient communication between each other. 2) Where are the files stored? The collaborative environment server has access to one or more FTP-accounts. It is possible to use free Webspace-Accounts! If files are important, they can be stored on several sites. Only the collaborative environment bothers with the details of storage-locations. No user of the Environments needs (or should have) access to the environments storage-places - they only need their user-id and password to the collaborative environment. Storageplaces can dynamically added and removed from the environment. 3) How do I search for files? The files are not simply stored "as is". The collaborative environment is some kind of distributed database management system. All files can have user-definable attributes. Files can have multiple revisions (Important if working with shared files!) Older revisions can be branched.. The use of user-definable attributes should be used to let people forget that they are working with files. They should think in a more "objectoriented" manner. Question: Would it be better to use a central-server for the collaborative envrionment like "ICQ" or should we use a more decentralized approach. (E.g. by using IRC to find the other participants and then create direct connections?) Jochen Schmidt [jsc--dataheaven--de]

 [2/7] from: rchristiansen:pop:isdfa:sei-it at: 22-Sep-2000 18:19


> Question: > Would it be better to use a central-server for the collaborative > envrionment like "ICQ" or should we use a more decentralized approach. > (E.g. by using IRC to find the other participants and then create direct > connections?)
Depends on how it would be used. If the application would be used as to create a sort of "ad hoc" collaborative environment, then the decentralized approach is appropriate. But if the application would be used to create planned environments, then a centralized approach is best. The application would be hosted by an Application Service Provider, which everyone with a big pipe wants to be nowadays. -Ryan

 [3/7] from: ryanc:iesco-dms at: 22-Sep-2000 17:07


For application proliferation purposes, not requiring a server would help out alot. [RChristiansen--pop--isdfa--sei-it--com] wrote:
> > Question: > > Would it be better to use a central-server for the collaborative
<<quoted lines omitted: 10>>
> wants to be nowadays. > -Ryan
-- Ryan Cole Programmer Analyst www.iesco-dms.com 707-468-5400 I am enough of an artist to draw freely upon my imagination. Imagination is more important than knowledge. Knowledge is limited. Imagination encircles the world. -Einstein

 [4/7] from: jsc:dataheaven at: 23-Sep-2000 4:30


Am Sam, 23 Sep 2000 schrieben Sie:
> For application proliferation purposes, not requiring a server would help > out alot.
Yes this is why I would prefer such a solution. I think I do not relly like the centralisitc approach. I also dislike conservative client/server paradigms. I would prefer a mobile-agent based solution. I had this idea some time ago: If I want something done, I sit down on my (dialup) computer. I'm offline and open my personal-agent communication program. My personal agent appears and I describe him the task he has to do. I go online and my agent transfers its corpus over the line to one of the 24h online agent runtime environments. Now I can go offline, while my agent e. g. watches the stocks (and warning me by SMS if something interesting happens) Perhaps a similar agent-based approach will be interesting for the collaborative environment. The whole collaborative environment coult consist of temporary and permanent online agents that communicate to each other. So one permanent Agent could serve the Environment by providing access to some storageplaces of its own. A temporary Agent could transfer a new file into the Virtual Space. Whats your opinion on this? Jochen

 [5/7] from: dynalt:dynalt at: 23-Sep-2000 8:36


Consider Code Co-op from Relisoft http://www.relisoft.com/ It is a distributed version control system that works using email. A similar approach using peer-to-peer connection would work nicely. Garold (Gary) L. Johnson

 [6/7] from: ryanc:iesco-dms at: 25-Sep-2000 10:55


Well, a permenent agent is from a perspective a server. Peer to Peer can also be viewed as client/server to client/server. Your permenent agent distinction does help me think in terms of who will be serving, say a group leader aka server. A series of backup leaders could be made possibly. My backup leaders idea is stinky, but may be workable. A problem I see that we must deal with is can we apply current standards such as IRC, NNTP, FTP, etc, with a non-specific server type arrangement. I dont think so. Therefore we must make either toss these protocols (yuck!), or adopt a leader/server style environment. I guess this means we must make it very easy for people to create a server, and deal with the server not being available. Maybe this is just an issue with NNTP? FTP sites are a dime a dozen, and I dont know how IRC works. Maybe we should just focus on the server end, most/all the OS's have the clients available. Make it a peice of cake for anybody to start thier own IRC, NNTP, FTP, Mail, mail list, and WWW server. --Ryan [jsc--dataheaven--de] wrote:
> Am Sam, 23 Sep 2000 schrieben Sie: > > For application proliferation purposes, not requiring a server would help
<<quoted lines omitted: 20>>
> Whats your opinion on this? > Jochen
-- Ryan Cole Programmer Analyst www.iesco-dms.com 707-468-5400 I am enough of an artist to draw freely upon my imagination. Imagination is more important than knowledge. Knowledge is limited. Imagination encircles the world. -Einstein

 [7/7] from: dynalt:dynalt at: 26-Sep-2000 19:10


Consider looking at the approach that Gnutella takes. It achieves an effect similar to Napster but without a central server. Perhaps a similar model with group restrictions or membership could serve as a base for at least some parts of the problem - file synch, etc.? I think the project is on SourceForge. The general approach to file synchronization taken by Code Co-op (a distributed version control system using email) www.relisoft.com is rather well explained. It essentially uses individual machines as servers for one or more projects - different projects have different 'homes'. Thanks, Gary

Notes
  • Quoted lines have been omitted from some messages.
    View the message alone to see the lines that have been omitted