Thanks for starting the Wiki page. This is all good food for thought and I'll try to collect mine and make a contribution.
-----Original Message----- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Alan Moore Sent: Wednesday, November 16, 2005 11:26 PM To: jess-users@sandia.gov Subject: Re: JESS: Debugger [EMAIL PROTECTED] wrote: > I think Jeffrey Davine wrote: > >>I'll do the research, but as a general matter, do you think the Jess API >>exposes enough information (particularly concerning the rete tokens) so that >>I could write such a tool? > > > Alan's more optimistic, but I myself don't think this can really be > done at the level you'd like without using and/or modifying private > APIs in Jess. The details of the Rete network implementation aren't > exposed as public APIs just because they change often and exposing > them would present too great a support burden. I was speaking strictly of my old WSH debugger code and utilizing Jess' implementation of the Eclipse debugging API. Of course these APIs are not considered public since they are in the gov.sandia namespace but they do implement public interfaces. However, that is far easier than modifying the Jess sources as I had to do way back when. The old code doesn't support viewing the Rete, etc. - it only does Funcall iterception for stepping through the RHS. I hope I didn't mislead anyone and am sorry if I did... > > But I'll tell you what: one enormously valuable contribution in this > area would just be writing use cases/storyboards/walkthroughs/UI > mock-ups. I personally have a hard time picturing just how this will > be integrated into the more traditional source-oriented Eclipse > debugger architecture. Remembering that the left-hand-sides of rules > can be shared in the Rete network, it would be extremely helpful for > someone to present a coherent view of how this kind of debugging would > work. I've put my suggestions in the wiki. Hope it helps. > > Personally I always imagine something like the existing Rete Network > View in the JessDE, but "live", connected to a debugged > program. Further, there's a notion of the "current Rete node" which is > highlighted in the diagram, and you can select an inspect the current > node or any other node. You can push a "step" button which somehow > shows you which tests pass and what Tokens get created, and you can > watch them go from that current node to the other nodes. But the > details are still fuzzy to me. How does this interact with stepping > through procedural parts of the program? This sounds reasonable. On the other hand, I think many Jess users would be more confortable if the complexity of the Rete is hidden or progressively revealed to them in a way that matches their cognitive view of their rules. Maybe a simplified way of drilling down via the LHS patterns into the (filtered) Rete state would be a place to start developing use cases. > > Can we start a Wiki page about this? ... OK, here you go: > > http://herzberg.ca.sandia.gov/jesswiki/view?ConstantineDebuggerPlanning > > Any and all contributions welcome. See you there! alan -------------------------------------------------------------------- To unsubscribe, send the words 'unsubscribe jess-users [EMAIL PROTECTED]' in the BODY of a message to [EMAIL PROTECTED], NOT to the list (use your own address!) List problems? Notify [EMAIL PROTECTED] -------------------------------------------------------------------- -------------------------------------------------------------------- To unsubscribe, send the words 'unsubscribe jess-users [EMAIL PROTECTED]' in the BODY of a message to [EMAIL PROTECTED], NOT to the list (use your own address!) List problems? Notify [EMAIL PROTECTED] --------------------------------------------------------------------
