Meeting
OHS Project , the unfinished revolution previous meeting, next meeting. 1 Agenda: 2
Eric Armstrong
Minutes: 4 Minutes of meeting held 4 - 6 p.m. in the SRI Engineering
Building, EK255.
Grant, Lee, and Eugene have posted diagrams from last week's meeting on the WIKI. 4B 2) SourceForge is up. We need to start organizing all of the info we've created over the past several months into a cohesive web site. Eugene suggested using our web site hosted by SourceForge (http://ohs.sourceforge.net/) as our developer's site. That way, we can use SourceForge's admin tools, and not have to bother the administrators of bootstrap.org. We can then create links from bootstrap.org to ohs.sourceforge.net. To edit the web page, you need to register on SourceForge, become a member of the OHS project, and then ssh onto the server. Eugene will mail out specific information on accessing the site to ohs-dev. 4C 3) Funding and participation. Doug reports that as the project has been progressing, every interaction about funding have been very positive. Eugene notes that the momentum has been good, and people have taken on projects. 4D Eric suggests we need a carrot -- give potential funders and incentive to invest in the project, a parameter that will drive investment. Eugene says that we should plan on having all of our information organizes, so that we can start approaching people with it in about two weeks. The carrot is different for different people and we've already got the carrots. For example, VA Linux has already seen the benefit; now they need to see some progress on our part before they will commit. Doug points out that the thing about carrots is that we do not want to give special privileges to a particular group, because we want this to be a worldwide thing. 4D1 4) Mailing list etiquette and task forces. Eugene suggests using the ohs-dev list for discussing all things pertaining to the development of the OHS, including non-technical issues such as process. Joe adds we should only post items directly relevant to the OHS on the ohs-dev list. Cool, interesting, general things not directly related to the OHS should be posted to the unrev-ii list. Eugene says that we should self moderate the list, and use our best judgement. Doug suggested that we should create multiple lists for each different domain of discourse. Eugene suggested waiting for now, until traffic starts to grow out of hand. 4E 5) Jack's presentation on onto-centric design. Doug and Jack met last night and discussed the idea of reverse engineering Augment. Developed a preliminary ontology for Augment, using versions of GraphMaker and Jext, two Java-based tools that Jack hacked up to do what he wants. 4F These tools allow you to graphically or textually develop ontologies, using an XML format. Jack intends to make these available on SourceForge. The resulting XML can be fed into an expert system to spit out UML in XMI (ArgoUML) format, which can be used as the basis to generate code. Jack calls this onto-centric design. 4F1 Jack also introduced an open source ontology tool developed at Stanford called Protege. Jack wants to add a DKR tab to this tool, so that its functionality is integrated to what we're doing. 4F2 Doug says the better Jack gets at this, the better we can use it to go to the next stage of the OHS. Have Howard or Jack to improve the ontology. Doug's intuition tells him this approach of using ontology will be basic to boosting collective IQ worldwide. 4F3 Jack says through evolutionary epistemology -- hack and try -- we'll get something going. Brian says he does not see in here how to express the user experience. Jack responds that this tool is not GUI-centric, it's algorithimic-centric. It's yet another way to do software projects. Howard has an ontology for OHS. Jack says he is mapping the topic map XML into his program, The Scholar's Companion. Doug says it would be good to have some interns working with Jack to get the ontology project going. 4F4 6) Architecture and views. Discussed the viewing engine. Doug briefly demonstrated the Augment SmallTalk client, and some of its navigation and view control features, which could easily be reimplemented in a browser. Brian talks about the Javascript part of the specs. The key question is, what are we going to build on. Doug says it has to be in an environment where it can evolve. Eugene notes we've determined that it's going to be HTML based. Brian says we need list of possible views. Eugene writes on the whiteboard the views that were brainstormed on July 25: 4G Message views for single discussion:
---
Above space serves to put hyperlinked
targets at the top of the window
|