[04:43:16] ** thinkeryzu has left IRC ("Download Gaim: http://gaim.sourceforge.net/") [04:43:35] ** thinkeryzu has joined us [05:24:36] [connected at Wed Sep 12 05:24:36 2007] [05:24:36] <> *** Looking up your hostname... [05:24:36] <> *** Checking ident [05:24:36] <> *** No identd (auth) response [05:24:36] <> *** Found your hostname [05:24:36] <> *** Your host is clarke.freenode.net[clarke.freenode.net/6667], running version hyperion-1.0.2b [05:24:38] [I have joined #peak] [10:20:06] ** m0no has joined us [11:26:33] ** pje has joined us [13:10:00] ** coderanger has left IRC () [13:10:13] ** coderanger has joined us [13:24:33] ** tav has joined us [13:24:40] heh, i never realised! [13:24:44] hello all! [13:26:57] * tav waves to coderanger / m0no / pje / thinkeryzu -- anyone around or just lurking? [13:27:08] I am around [13:27:12] * pje is here [13:27:17] though only briefly [13:27:29] never realized what? [13:27:30] pje usually will respond pretty quickly [13:27:32] heh [13:27:42] As long as you speak my name, anyway. [13:27:48] that there was a peak irc channel [13:27:52] Ah [13:28:16] so, how goes trellis? i noticed it quietened down on the list front [13:29:07] I'm working on the revised algorithm and "universal co-ordinator" [13:29:10] should be a post on it soon [13:29:43] I'm getting near to a co-ordination strategy that would allow for pluggable things like undo, threading, rollback [13:29:47] "universal co-ordinator" ? [13:29:56] Virtually a Python STM implementation [13:30:07] nice [13:30:39] one can then finally throwaway the zodb =) [13:30:41] See http://www.eby-sarna.com/pipermail/peak/2007-August/002782.html for the first draft [13:30:52] Er, STM isn't persistence [13:31:01] Though you could build persistence on it. [13:31:21] i realise -- but, the zodb is the closest thing to stm in the python world i find [13:31:50] I suppose -- except that it can't do threading/pessimistic locking [13:32:18] the co-ordinator concept I have in mind would actually allow objects that share a co-ordinator to be shared across threads, using the co-ordinator as the locker. [13:32:29] Assuming the co-ordinator supported that, that is. [13:34:16] nice [13:34:36] is there an eta on implementation? and what can others do to help? [13:36:15] ^ pje [13:36:56] er, no to the first, dunno to the second [13:37:06] Maybe "give me good ideas?" :) [13:37:54] well, trellis is v.similar to thoughts i'd had on disco -- http://wiki.espians.com/Disco [13:39:30] unfortunately, the implementation https://svn.espnow.net/24weeks/trunk/source/python/disco/base/disco.py ran into some problems i didn't have time to resolve [13:39:38] and trellis seems to be resolving some of them [13:42:21] but, there are some things you're not tackling, e.g. capability-based access, &c. [13:42:55] have you had a look at http://www.cypherpunks.to/erights/talks/thesis/markm-thesis.pdf ? [13:43:49] it takes a unified view on "object-orientation" + access control + concurrency control [13:45:25] scope of the trellis work is determined by client needs [13:45:32] i.e. OSAF and Chandler [13:46:07] pity cos you're doing some amazing work here [13:46:46] how are they managing access control at the moment ? [13:49:00] What access control? [13:49:15] The server stuff is all in Java [13:49:21] different system [13:49:40] ah [13:52:36] so, failing at being helpful, can one at least peruse the unreleased work-in-progress of trellis? [13:54:42] no :) [13:54:53] There's the *released* work that's finished, though. [13:55:03] Including a post-0.5b1 bugfix, I believe. [13:55:05] gah! ;p [13:55:26] won't the api change rather dramatically? [13:56:02] i'm looking at building an entire web framework (not that the rest of peak isn't fine) on top of trellis [13:57:22] Not the main API, I don't think. But the part for creating custom data structures will. [16:30:11] ** bitmonk has joined us [20:04:14] hmz [20:36:07] ** pje has left IRC ("Client exiting") [21:31:01] ** bitmonk has left IRC ()