From: Adam R. <Ad...@Kn...> - 2002-12-12 08:35:37
|
Day: 34 # of developers: 10 # of users on this mailing list: 20 # of mailing list posts: 31 # of downloads: 62 # of applications checked in: 38 # of Web hits: 1463 Overall ranking: 5961 out of 52,503 Joyce wrote: > Hi all, I've managed to get a public mod-pubsub server up at: >=20 > http://mod-pubsub.thrillerguide.com >=20 > Feel free to play on it, give demos, whatever. Also note how slow it > is. This is a reasonable server at a decent data center with a fatty > pipe, so I think with the current code (untuned) this is about as > good as it gets. JP Thanks! Which chat room in there should we use for real-time chat among each other? Elias wrote: > Gah, deja vu! I've been thinking of starting a mod_pubsub project for = a=20 > little while now but really had very little idea of where to start. = The=20 > development of event notification for the web is a little daunting,=20 > given the strong oppinions that are out there about how it should be=20 > done. I just sent Adam an email asking to be added to the developers=20 > group on sourceforge... Very exciting! Will do... Tom wrote: > The key thing is all the users would need to play is a browser that = can > deal with mod_pubsub. Amen, brother, you're singing my Nep Tune. I do have live web browser versions of Connect Four and Reversi that are essentially hacked up from their originals from http://www.irt.org/games/ Should I clean those apps up and check those into the distribution as = well? They never worked quite right because Rohit took them 80% of the way = there and then lost interest. Basically, they need some added start-game and coordination work. Mike wrote: > So mod_pubsub isn't the center, its the in-between. Very well stated. We need to write up a philosophy page on the REST = Wiki or something... mod_pubsub does not have any way to compute new state. It just transfers the state to the listeners. Derek wrote: > Hi, heh, kinda like a high-school reunion (hi Kragen, Ben, Rohit, = Adam!) > or maybe like one of those 12-steps recovery groups ("and then they = gave > us $10 million ..."), anyway real nice to see you all again -- Nice to see you, too! > Been doing some Jabber scripting lately, scenario's an in-browser = MMPOLG > w/ in-game JSRS-type chat, the Jabber-bot would allow out-of-game = players > to participate via IM clients (AIM, Yahoo, MSN etc.) by proxying = messages > to and from in-and-out-of-game players & groups ('chatrooms') through = the > J-bot. Before settlng on Jabber there'd been some speculation about = maybe > using KnowNow's pub-sub messaging framework for this ... then I saw = Scott > Andrew's heads-up fly-by a few weeks ago ... then my hard-drive died = ... > anyway the Jabber stuff's cooking just fine, but there's still this = open > question about possibly integrating KN messaging w/ Jabber -- ??? No question. It *can* be done, and it *will* be done. It's just a = question of who has time to take this one first. > Anyone have old ideas on Jabber/KN hybrids they'd care to dust off & = take > for a brisk morning trot? Ask Rohit, he's always had a soft spot in his heart for Jabber. Fred wrote: > An example of what KN should be able to do. >=20 > http://www4.passur.com/sjc.html > Tracks all air traffic in near-realtime in the San Jose area. Appears=20 > to track anyone with a mode C transmitter, so you can watch your=20 > friends in the pattern at Reid-Hillview or PA. Hey, cool find! Completely off topic, how do we get an entry at http://www.advogato.org/ = ? Fred wrote: > [Live chess] requires an older version of the C Microserver (the one I > wrote). Don't know if we're going to get that source code. You will, it's just a matter of time. Product management at KnowNow has been asking for libkn packaged as a Microserver SDK for at least 6 = months; we just gotta find some engineering cycles to package and release it. Other things on the to-do list that folks on mod-pubsub-developer can do if any of you have time: 1. Make a PHP Microserver using SWIG creatively. 2. Pull Phil Harris' Python Microserver into the distribution. 3. Submit PubSub perl package to CPAN. 4. PubSub-enable SOURCEFORGE so we can see stats as they happen = instead of (ugh) many days later. 5. Demonstrate *asynchronous* XML-RPC. (Woo hoo!) 6. Invite folks from Pushlets, XMLBlaster, Jabber, or OpenThought = projects. 7. Anything on the TODO list: = http://cvs.sourceforge.net/cgi-bin/viewcvs.cgi/mod-pubsub/mod_pubsub/TODO= Here's that TODO list in the cvs repository in case you're too lazy to = read it... ------------------------------------- High-level items on the "to do" list. ------------------------------------- 0. Instruct Apache to redirect /kn to pubsub.cgi . 1. Create a mod_pubsub self-contained Apache module that subsumes the best practices of pubsub.cgi and pubsub.py . 2. Create a php_pubsub client. 3. Revamp RSS Support because kn_sense/rss is an awful hack. Use Jeff Barr's wonderful http://www.syndic8.com/ tools. 4. Create a PubSub::Server.Pm and make pubsub.cgi a thin wrapper around it. (Note: we have already merged "KNMS" into Client.pm .) Part of this project may include making Tunnel.pm event-driven. 5. Unify the Perl PubSub client and server and submit the PubSub perl package to CPAN. 6. Import (and extend) Phil Harris' python pubsub client from http://cvs.developer.knownow.com/ 7. Clean up python_pubsub so it doesn't need to ship with all those other .py files. (Ka-Ping Yee? We wish...) 8. Get push_manager working and clean up its directory. 9. pubsub.cgi should be less lazy about detecting closed sessions. Bring active presence detection to mod_pubsub. 10. While we're at it, clean up any broken and/or ugly kn_apps. 11. pubsub.js -- fix to deal with long lived IE6 sessions. 12. Build an Apache log sensor and use that to create a "webmaster's dashboard" application. 13. Get "Pocket NOC" application that monitors machines working to create a "sysadmin's dashboard" application. 14. Write a Jabber gateway using mod_pubsub. 15. Write an (all-DHTML) IM client application using mod_pubsub. 16. Get Pushlets working with mod_pubsub. 17. Import applications from http://cvs.developer.knownow.com/ 18. Improve and extend pubsub_test.cgi . 19. Build performance testing tools. 20. Import Greg Burd's ruby pubsub client from http://cvs.developer.knownow.com/ (Didn't think we were going to forget that one, did you? :) 21. Get working Fred Sanchez's chess application from http://cvs.developer.knownow.com/ 22. Start a Wiki for this project. (?As part of the REST Wiki? Talk with Mark Baker & Jeff Bone...) 23. In the kn_apps directory, unify kn_lib and jscomponents. ---- Ad...@Kn... I'm glad to see KnowNow re-embracing open source by releasing some of their application server technology. It appears to be the same code that wowed people almost two years ago at the first big KnowNow demo, and it's good to see the company get back to its roots. Back when I was a part of the company, I (and others on the web app team) sat in meeting after meeting in total disbelief as plans were made to abandon all developer relations, to stop being an open-source friendly outfit, and instead become an enterprise software corporation. Why they chose to abandon all the great buzz and community support they had in March of 2001 just to try turning a big buck (instead of free code, price tags in the tens of thousands were stuck to everything) on unproven software should be a lesson for MBA courses everywhere. Market buzz is about the best momentum you can have in a new startup, and doing everything you can to kill it is giving yourself the kiss of death. Mark my works, once someone ports the code to an Apache module you'll start seeing some wicked-cool applications using it. Heck, I'd finally make the switch to Apache just to use it on MetaFilter (real-time message counts/comment output/instant messaging). -- Matt Haughey, = http://a.wholelottanothing.org/archived.blah/12/01/2002#1065 Bill Gates said the way to make money is by setting de facto=20 standards and implicit in this is to first create the standard in=20 such a way that it can become de facto. The enterprise software=20 market works the other way, trying to trap customers and suck the=20 blood out of them. I agree with the author. -- Robert X. Cringely, in response to the above passage |