[18:47] trike still on for tonght? [19:01] yup! [19:01] see you here at 7:30 [20:43] oki! [21:11] Action: Dymaxion looks around [21:13] wee [21:15] how's the day, dym? [21:18] ok [21:18] have you talked to brenda tonight? [21:19] yesterday, not tonight, though [21:21] Action: Dymaxion finishes dinner [21:21] yay! [21:21] Did you get your report done? [21:21] hiya! [21:21] yes, i did. [21:22] it wasn't even bad [21:22] woo! [21:22] ar: how was your day? [21:22] it was there ^^ [21:22] I feel like I accomplished oodles [21:23] and nothing at all :) [21:23] Action: Dymaxion grins [21:23] asparagi: was that your big report? [21:24] Action: Dymaxion blinks [21:25] I just got my new flash disk in the mail, and it came with some crufty software on it, so the first thing I did was to format it. [21:25] It took < 2 seconds to format a 2GB drive. [21:26] wow [21:26] Action: Dymaxion nods [21:26] I knew it was supposed to be fast and all, but [21:26] :) [21:27] oh, btw...congrats guys on the v1 adoption :) [21:27] that was awesome to see ^^ [21:28] ok, sorry, i had to get food. now i'm really here [21:28] i had 2 big reports. this was the 2nd one [21:28] now my life doesn't suck [21:29] yay, for !sucking [21:29] so in the mails that passed about the v1 stuffs, it was suggested that we start convos w/ the ACE team [21:29] and also that we post v1's limitations to the site [21:30] with some hopeful tidbit that corrections are pending at some undefined future point [21:30] yes [21:30] i'm cool if someone else starts the convo w/ ACE but i don't want to talk to them personally [21:30] everything else I agree with, except for the fact that if we're fixing bugs that are in the v1 branch anyway, I don't see how it would hurt us to push just a fixer release [21:31] because we've already agreed to only fix bugs that traverse into v2 [21:31] I haven't had time to call Akshay; I'll try to do that tomorrow. [21:31] akshay rox ^^ [21:31] who else is on the ace team? [21:31] I'm also meeting with J. D. Meier on unrelated business tomorrow; I doubt trike will come up, but I'll let you know if it does. [21:32] Talah and Ahmed are the people we need to talk to, Sean to a slightly smaller degree [21:32] Neither Akshay or J.D. are on ACE [21:32] ah, gotcha [21:32] Action: ar4chne isn't sure what exactly the ace team is [21:33] oh, who's akshay working for, then? [21:33] JD is SWI, Akshay is MSC [21:33] MS consulting [21:33] ahh [21:33] Action: ar4chne doesn't know mucch about ms internals [21:33] except I have a few other friends on the swi team [21:33] Action: asparagi wishes she knew less [21:34] ACE team is a group that does security work for line of business applications -- internal stuff inside MS, basically. [21:34] Oops. :-) [21:35] well, about pushing v1 releases, let's see when we have code in hand [21:35] Sounds good [21:35] mostly i don't want us to lose our bleeding edge rep [21:35] we have a rep? [21:35] Action: Dymaxion grins [21:35] :) [21:36] yah, we're an industry-standard threat modeling methodology [21:36] didn't you get the memo? ;) [21:36] Action: Dymaxion . o O ( God help this industry ) [21:36] heehee [21:36] Action: asparagi laughs [21:36] k. so, posting v1 limits to the site [21:38] we should probably write it up a llttle more publically-consumable than paul already sent out [21:38] Yeah [21:39] writing doesn't really feel like a good group project [21:40] I'll take a stab at it [21:40] i think maybe we have a weirdness in our work flow. we have allocated all our Trike time for meetings when only part of our task list benefits from synchrony [21:40] awesome [21:40] oki, then what else is on the agenhda? [21:40] i'll try not to get too meta [21:41] well, is this the theory meeting or the tool meeting? [21:41] This is the tool meeting [21:41] tool [21:41] ok. [21:41] so let's do a little tool planning [21:42] like, we should write the canonical list of support libraries we know we need [21:42] then we can identify candidates, or write them or whatever [21:42] Action: Dymaxion nods [21:42] also, we should get synced up a little on what this puppy is going to look like [21:43] that should help identify what ui library support we need [21:43] so, I've been thinking [21:43] and it seems that we're going to end up doing a lot calculations / comparisons [21:43] right? [21:43] and that seems to be what slowed us down w/ the v1 tool, wasn't it? [21:43] i think we should go for a little while working on components & let the theory get further before we settle on feature list for v2.0.0 [21:43] yes [21:44] I vote we export some stuff to clibs [21:44] for speed [21:44] hmm [21:44] i think that particular problem was a stupidity problem vs. a Squeak problem [21:44] I think we should do that when/if we've found real speed problems and exhausted our easy profiling options [21:45] the algorithm was very inefficient & it was being invoked many unnecessary times [21:45] Having everything in squeak is a big deal while we're still in code-churn [21:46] oki [21:46] i could see exporting some stuff to C at some point... i'm not totally against it. but yah, if we let it float a little that'd be nice imo [21:47] toolwise, i think our major needed libraries are: [21:47] - SOAP [21:48] - XML/XSD [21:48] - gesture recognition [21:48] - structured drawing [21:48] - tabbed UI framework [21:48] does that sound like it? [21:49] That's everything i know about [21:49] Well, graphs [21:49] yah... graphs [21:49] ara, can we mind meld you about the UI? [21:49] how so? [21:50] its pretty straightforward, no? [21:51] well, way before v1 we were talking about a very pen-friendly, diagram-based UI [21:51] yeah, you guys've mentioned it [21:51] ok.. [21:51] the only thing I would caution is that not everyone has pens :) [21:51] so I think we should have drag / dropable drawing components [21:51] Yeah; we want it to be mouseable as well [21:51] yes, it has to be keyboard & mouse friendly too [21:52] so the way it should feel is that you're just sketching notes on paper, except the paper is smart & sort of carries your ideas forward [21:52] omfg...I'm going to kill my cat [21:53] it should let you do things that are inconsistent, but flag them as inconsistent [21:53] :( she just ruined one of my favorite scarves pulling the sequins off [21:53] suck! [21:53] how it has holes :( [21:53] :( [21:53] bad kitey [21:53] er, kitty [21:53] suck! [21:54] i think it might be beneficial to try getting the UI up sooner in the process this time [21:54] i think we are well-positioned to do that because we can reuse stuff from v1 [21:54] Yeah, that doesn't sound like a horrible idea [21:55] A lot of our bug fix delay has been for UI stuff, too [21:55] yah. it'd be good to get it some more run time [21:57] hehe, I feel like we're a little unfocused atm [21:57] yup [21:57] sorr [21:57] so, basic components [21:57] i was trying to figure out how we should approach the libraries [21:57] some sort of workspace for the drawing [21:57] yes [21:58] some sort of dragable widget thing that gets converted to a drawing component [21:58] still need better balloons for the balloon help (; [21:58] Action: Dymaxion grins. [21:58] Is there anything on the library list that we don't have a strong candidate for? [21:58] we need to do some searching for BookMorph clones & see whether we want to keep Spiral, the one I wrote, or whether someone else has a better one for our purposes [21:59] Ok [21:59] the draggable widget thing will come from Connectors, which is a gorgeous library for structured drawing [21:59] If there are docs, I can offer my opinion, but I doubt there are [22:00] here's what I translated our list into: [22:00] Graphs SOAP on Flow XSD Class Factory Genie Connectors stick with Spiral or combine w/ other BookMorph reimplementation? [22:00] erg. [22:00] damn newlines [22:00] what is SOAP? [22:00] Graphs [22:00] SOAP on Flow [22:00] XSD Class Factory [22:00] Genie [22:00] Connectors [22:00] stick with Spiral or combine w/ other BookMorph reimplementation? [22:00] do you mean like simple object access protocol, or sthing else? [22:00] yeah, that [22:00] yes, [22:00] ok [22:01] That's how we're looking at having stuff talk to trike [22:01] ah [22:01] eww [22:01] k [22:01] so there's a SOAP implementation, but it's ass [22:01] do you have other protocol ideas? [22:01] i would be happy to swit ch [22:01] definitely [22:01] More than happy, even [22:01] esp. since as i said, the existing SOAP implementation is ass [22:02] hmmm...lemme research a bit [22:02] so we will need to write tne [22:02] :) so that I can give an intelligent response [22:02] instead of just 'ick' [22:02] there is a decent XML-RPC implementation [22:02] Action: Dymaxion nods [22:02] because all my experience w/ soap before has been 'ick' [22:02] awesome. you are in charge of networking protocol [22:02] Why were we looking at soap instead of xmlrpc? [22:02] (just curious, I can't rememeber) [22:03] hmmm...xml-rpc wouldn't be bad; since we already have our xml structure put together [22:03] i think it was more full-featured & better supported across dev env [22:03] because it's all trendy now [22:03] Huh, ok [22:03] google does it! everyone should!@#$! [22:03] xml-rpc would be way easier, i think it was mike who wanted soap [22:03] Action: Dymaxion throws crockery [22:03] Action: Dymaxion nods [22:03] I vote easy [22:03] google uses soap? [22:03] maybe we should post a question to mike? [22:04] Let people who want to interop do the work [22:04] http://www.google.com/apis/ [22:04] i mean, so do i every morning, but.. [22:04] ;) [22:04] hehe [22:04] hmm...anyone know what wsdl is? [22:04] yah, let's post our intention to use xml-rpc & ask mike to make any violent objections now [22:05] web services discover language [22:05] it's a protocol for exchanging information about the protocol you support [22:05] er, discovery [22:05] ah, gotcha [22:05] it's how you find out about a soap interface [22:05] i almost said soup interface [22:06] Close enough [22:06] if someone else makes a list of bookmorph clones or other leads on tabbed interfaces, i can take a stab at evaluating them [22:06] Action: Dymaxion . o O ( throw in fourty outsourced developers, stir for a six week product cycle, serve chilled to an uninterested public ) [22:06] Where should I start looking? [22:07] and what should I be doing? [22:07] besides twiddling? [22:07] the squeak swiki? google? [22:07] ok. I'll poke [22:07] ooh, we should codename v2 to 'twiddles' [22:07] :) [22:07] well, lessee [22:07] and we can be all "yes, twiddles is coming along nicely" [22:07] hehe [22:08] an important task is to gather all of these libraries & try to get them all installed into the same recent squeak image [22:09] it sounds kind of awful, but if you want to work with me on that we could probably get your squeak SkillZ ramped right up [22:09] oh, another option would be proprietary header, ja? [22:09] the other alternative is probably dev work carrying the half-written XSD Class Factory forward [22:09] proprietary header? [22:10] vs. twiddles? [22:10] or XML-RPC? [22:10] or ? [22:10] proprietary header vs. twiddles, fight! [22:11] ok, here's a relevant question [22:11] do we want a stateful or stateless protocol? [22:11] what exactly is the point of the communication? [22:11] stateless [22:11] we will have a server with the canonical version of the model & one or more clients sending it changes [22:12] it should send updates to the clients as the model changes [22:12] Will we ever need transactions? [22:12] oh. [22:12] yes, probably [22:12] Do we want to do them ourselves or have the protocol handle them? [22:12] so we're going to need some sort of pki? [22:12] e.g. i completely disassemble the model but don't want others to see the intermediate stages [22:12] eeeeee [22:13] no, transactions like database transactions [22:13] pki [22:13] we'll need an auth model... god I hope we don't need pki [22:13] that is a good point, we need some seriously configurable auth [22:13] add auth library to the list [22:13] i don't really want to deal with transactions but i don't see how we can avoid it [22:14] we don't want to write that [22:14] is there a squeak corba orb? [22:14] fifty million ton hammer, but. [22:14] i dunno [22:14] Might be another option [22:15] It'll make the C++ people happy, fsvo happy, but it'll make life painful for everyone else. OTOH, it may get us things like transactions for free [22:15] it sounds kind of heavyweight, i.e. if there is an implementation it will be spotty [22:15] yeah [22:15] well, it seems like what we put in a transaction is seriously app-dependent [22:16] yeah [22:16] so how could an external library help? [22:16] Well, it provides support for transactions as a concept very easily, etc. [22:16] i see [22:17] It may even just let us declare the model as synchronized objects or something, and then just wave a magic wand [22:17] oh. squeakmap is another place to look for bookmorph clones [22:17] I've never gotten too close to it, though [22:17] eee [22:17] ok, what would a transaction consist of? [22:17] it sounds sharp [22:17] like, we want the ability to rollback changes? [22:17] several changes to the model at once [22:17] probably [22:17] oh yeah [22:17] definitely [22:18] deleting an actor, say, and fixing up all the rules to make things pretty. [22:18] yah [22:18] so semaphores / mutexes or some locking mechanism? [22:18] Yeah [22:18] altho there's also the ? of how smart we should expect the client to be [22:18] not very [22:19] maybe the client should be totally dumb and when you delete the actor, the server fixes all the rules automagically & sends everybody update notices [22:19] hrm. [22:19] i guess it needs to be both push & pull [22:19] what do y'all think about just using SSL certs for mutual auth? [22:20] Definitely not a bad start; I think that's probably good enough, really. [22:20] well, I def think we should ssl the comm [22:20] Auth is a nice-to-have that we can work on once we've shipped 2. [22:20] because otherwise you'll be sending info about vulns all over [22:20] in readable format [22:20] i think we should have it from the beginning [22:20] but you're still going to need an auth mechanism [22:20] oh, we have to have some kind of auth [22:21] because otherwise anyone can make changes [22:21] I meant supporting more complicated auth stuff than basic ssl certs [22:21] oh, ok [22:21] so, SSL for authentication but we still need authorization on top of that [22:21] i'm thinking per-model, read & write [22:22] yeah [22:22] do we need something more granular than that? [22:22] I don't think so [22:22] models are too integrated for it to be very meaningful [22:22] mmm, yah, ok [22:23] Action: Dymaxion fends off a cat [22:24] If we ever get composite models working, we might revisit that question, but I'm not too inclined too [22:24] wow, my cats are being little angels tonight [22:24] Action: Dymaxion grins. [22:24] This one just wants a lap. [22:24] must be the kitten [22:24] nope [22:25] Ok, so what libraries need identification other than network and UI? [22:26] well, now we have SSL to deal with [22:26] last I checked there wasn't an implementation [22:26] ok, so do we think we're going to need complex data to push / pull ot the server? [22:26] yes [22:27] right; we'll need to figure out using an external implementation [22:27] maybe an openssl plugin [22:27] ? [22:27] Action: Dymaxion nods [22:28] you guys still think that the pretty graphix lib is going to be worth the headache we're about to embark upon with the auth / xml stuff? [22:28] so, the complicated content will tend to come from the server to the client [22:28] Action: asparagi shrugs [22:28] like...in c it's a simple problem...been solved many times over [22:28] https://lists.wisc.edu/read/messages?id=183375 [22:29] what would we need to do to get a cross-platform structured drawing editor which does gesture recognition? [22:29] huh [22:29] or we could just use someone else's [22:29] and for osx / linux? [22:29] of course i don't exactly want it as a dll [22:30] ar: working on it [22:31] http://people.squeakfoundation.org/proj/msh-crypto [22:32] wow, and it even mentions spoon [22:32] it has to be an active project [22:32] Action: Dymaxion nods [22:32] hey, and i'd lay odds that msh, whose initials i've never seen before, is Matt Hamrick [22:32] do we trust him to write crypto? [22:33] ooh, i'm right. [22:33] I don't think I know him [22:34] i have met him a couple of times. i am under the impression based on the reactions of the people i was with that he is a Name [22:35] cool [22:36] Action: Dymaxion doesn't have the appropriate plugin to be able to actually view the project page [22:36] it's the squeak plugip [22:36] yeah [22:36] i'm frightened [22:36] :) [22:37] ok, so not to be cranky...but .net is relatively xplatform [22:37] we _could_ switch to c# [22:37] So, it looks like we do or may soon have options for ssl [22:37] and wxwidgets doesn't have a horrible drawing lib [22:37] and it has a gesture recognizer and structured drawing library? [22:38] ...and the debugging tools we've been using? [22:38] that's not what you were saying when you were fighting with it for captain... [22:38] er, rather, something equivalent [22:38] i don't think i really care, at this moment [22:38] yes, but I've seen us fighting squeak bugs too [22:39] i don't think i'm too likely to write any code in .NET, but that's not a particularly good reason for the project not to use it [22:39] 3j's coming [22:39] e? [22:39] er, er? [22:40] :) [22:40] you're psychic.. [22:40] <_3ricj> fucking irc server doesn't like '3ricj' as a handle. [22:40] heh [22:40] so, who would write trike if it were in C#? show of hands [22:40] can't start with a letter [22:41] s/letter/number/ [22:41] I think [22:41] asparagi: don't be angry [22:41] no, i'm not. [22:41] is there no other language you'd consider? [22:41] it was a serious question [22:41] lisp would be next on my list [22:41] but it's not that high at this point [22:42] python, maybe [22:42] I'd be really torn. I've really loathed the c# dev work I've done. I'd be more inclined to go off and try to write it in python [22:42] <_3ricj> I'd be very tempted to write it in ruby on rails. [22:42] <_3ricj> but I'm just a tourist. [22:42] isn't that a web thing? [22:42] <_3ricj> yup [22:42] heh [22:42] I've never done python [22:43] i don't think the web can do what we need it to.. [22:43] OCaml? [22:43] haven't looked at it [22:43] i'd at least look [22:43] <_3ricj> crap, I join my first #trike meeting, and ya'll are talking about what to write it in!?!?!?!?! holly fuck people. [22:43] 3j, watch the logs, plz [22:43] :P [22:43] what did you want us to be talking about? [22:43] I'd think about OCaml, but the UI support isn't there [22:44] <_3ricj> asparagi, are you getting my msgs? [22:44] I dunno, guys...c/c++ or c# would be good in a lot of ways [22:44] nope [22:44] i don't think wxWidgets gets us remotely close to what i wanted to see for the ui [22:44] If you're not a registered user, you can't msg people by default here [22:45] They had a nasty spambot problem for a while [22:45] i could probably cope with C [22:45] <_3ricj> neat. an IRC server where you can't message people. [22:45] And just write out own UI library? [22:45] hmm [22:45] then why not go c++ and have it be 1/2 there? [22:46] i forgot about that momentarily [22:46] <_3ricj> before I duck out: I can't do 10pm asparagi, as I'm watching a film with pablos. Gets out at 11pm. [22:46] ok [22:46] i'll be asleep. try me tomorrow. happy film [22:47] ok. let's look around for some cross-platform structured drawing libraries & gesture recognition [22:47] c# has really good tablet and drawing support [22:48] how much of that is avilable cross platform? [22:48] does it do structured drawing, or just drawing? [22:48] no idea [22:48] <_3ricj> would it be bad and/or unwelcomed if I voice some of my thoughts on the matter? [22:49] there's all the ink stuff, but I don't know about anything other than visio to get structured drawing [22:49] no [22:49] I do know that wxwidgets has support for drag and drop [22:49] and iirc, the pen support is really just a wrapper for mouse events [22:49] so it's not hard to implement [22:49] go ahead, 3j, we're just talking... more thoughts are fine [22:50] http://www.manageability.org/blog/stuff/open-source-structured-graphics-libraries-in-java/view [22:51] GEF sounds sort of like Connectors. This is the kind of thing I am looking for [22:51] so what about *cringe* java? [22:51] I hate it as a lang, but at least I know it [22:51] and feel I could contribute more [22:52] http://www.nwoods.com/ [22:53] commercial [22:53] oh. [22:53] yeah, very not free, it looks like [22:53] that's funny, it didn't even occur to me to check for that. i forgot that not all software is free [22:53] maybe i am spacy tonight [22:53]