[00:05:42] ** lyrrad has joined us [00:40:46] ** gpciceri has joined us [00:52:22] ** lyrrad has left IRC ("Leaving") [00:53:13] hazmat is now known as haz|away [02:07:23] ** haz|away has left IRC (Read error: 60 (Operation timed out)) [02:56:29] ** gpciceri has left IRC ("Client exiting") [03:04:58] ** gpciceri has joined us [03:48:10] ** gpciceri has left IRC ("Client exiting") [05:12:32] morning [05:12:35] jack-e|away is now known as jack-e [05:18:57] ** jrobiez has joined us [05:21:45] ** jrobiez has left us [06:37:06] ** vlado has joined us [07:47:16] ** vlado has left IRC (Read error: 104 (Connection reset by peer)) [07:47:47] ** vlado has joined us [08:12:18] ** vlado has left IRC ("Leaving") [09:07:54] ** gpciceri has joined us [09:08:01] ** gpciceri has left IRC (Read error: 104 (Connection reset by peer)) [10:30:12] ** haz|away has joined us [11:07:29] ** redLED has joined us [11:42:45] nice EW [11:42:50] err WE [11:43:00] ** jack-e has left IRC ("(null)") [11:50:56] haz|away is now known as hazmat [12:16:46] ** pje has joined us [12:16:54] Hola. [12:17:17] afternoon [12:18:07] a'noon! [12:21:39] * pje is writing stored procedure tests with DDT for a porting project at work [12:21:56] Using MS Word, which so far has been a mixed experience at best. [12:22:08] I think next time I'll use reStructuredText. [12:28:10] *shudder* [12:28:15] word is pretty evil [12:28:53] *shrug* It's a proof-of-concept. [12:29:18] This didn't really need to be in Word, but I wanted to show that Word docs would work for this sort of thing. [12:30:47] The DB folks like it a lot, so far. :) [12:31:06] They keep talking about how they wish they'd had it for so many other projects. [12:31:09] I'm sure wordperfect documents would work too.. but, I wouldn't do it :) [12:31:14] Heh. [12:31:20] building some sort of a knowledge management system orso? [12:31:25] Well, Word is pretty much the standard thing around here. [12:31:47] redLED, no... I'm using peak.ddt to do document-driven testing. [12:32:01] Sort of an "executable test plan" [12:32:06] document-driven testing... *tries to visualise* [12:32:26] You write a document that describes what you're testing. You put sample data in tables. [12:32:45] In the first cell of the first row of each table, you put text that describes what kind of test it is. [12:33:07] And then you set up a configuration file that maps those descriptions to classes or expressions that use peak.ddt stuff. [12:33:12] *poits* [12:33:17] would'nt that me more the domain of excel? [12:33:21] You can then use the DDT tools to display a web page... [12:33:33] That highlights good and bad tests in green and red. [12:33:40] You can use Excel too. [12:33:57] The point is that you generate HTML, and then that becomes a "live" test script. [12:34:08] well, you just answered my biggest question since i came here :D [12:34:09] If you edit the file and save it again, you've updated your test data. [12:34:16] Reloading the page in your browser re-runs the tests. [12:35:00] So, as I add tests to the document I'm writing, in literally real-time the DBA folks can see what's wrong with their Oracle port of the Sybase procedures I'm writing the tests for. [12:35:18] And as they fix stuff, they hit reload to see if the red cells go green. [12:35:29] Meanwhile, I'm sitting here adding more tests. :) [12:35:35] ok :) [12:35:52] thats one of the feakiest things done with technology i have heard about in the past 24 hours [12:36:00] but i suppose someone pays you for it, and probably pretty decent, too :D [12:36:02] Feakiest? [12:36:10] Maybe you mean PEAKiest. :) [12:36:20] well, thats the question you answered me :) [12:36:27] see, i came here a while ago after reading PEAK's website [12:36:53] and by the time i finished, i went like 'so, is this like a very convinient way to abstractize development, or a bunch of words which vaguely describe some strange middleware' [12:37:16] now i'm starting to get some faint clue about where it actually comes in handy :) [12:37:35] DDT is one of the newest tools in the toolkit. [12:38:32] * redLED aims to eventually lead a small team of programmers on the creation of a self-contained and dynamically expansive, testable, and reconfigurable tool for large-scale network management [12:38:36] It works nicely out-of-the-box with the N2 tool; with N2 you can dump a database's contents as HTML, then with a little work you can run DDT on the results, to verify a database's contents. [12:39:08] basically, my aim is to feed my application a SNMP MIB, and have it figure out what the device is, what it does, and how it's existance should affect the rest of the network's operational policy [12:39:11] Interesting. I think JPL (John Landahl, not Jet Propulstion Labs) is working on something like that. [12:39:28] Well, maybe not quite like that. Dunno if SNMP is on his agenda. [12:39:49] SNMP is just one of many, serves a good example [12:40:43] but basically, i need to reach a stage where i can throw a device at a network which allready has the management application and some operational policy, and the device will be configured and maintained from there on, funcionally and administratively, with as little human interference as possible [12:41:38] * pje nods [12:43:37] what i'm considering (or atleast, attempting to understand in order to consider) is weither PEAK would be a good infrastructure to build on considering the very much needed abstractization and bastardization of seperate components; as in, could i eventually just toss in a python interface for 'yet another device', and have it magically become a part of my existing code infrastructure [12:44:07] and so far, my research led me to PEAK being the only thing which actually aims to do something of the sort :) [12:44:44] (java and the various beanies would be first consideation, but it's out of the question for me due to sun/licensing terms/uncertain future/etc.) [12:46:49] Java is also very heavyweight for that sort of thing, even using the new generation of "ultralight" containers. [12:47:07] (E.g. picocontainer, nanocontainer, etc.) [12:47:12] it doesent read nice, either. [12:47:17] thats why i love python [12:47:22] it's all very... clear somehow. [12:47:58] java is very explicit about everything, some love it, some don't [12:48:04] i prefer short, explicit, and readible [12:48:14] implicit, even [12:50:46] hey all [12:50:48] pje, how did the project installer sprint before pycon pan out? [12:50:54] s/project/package [12:52:02] It's not till the 22nd. [12:52:02] Or thereabouts. [12:52:27] There's some stuff up on the PyCon wiki, and the beginnings of some stuff in the PEAK CVS. [12:52:41] Specifically, I'm re-orging PEAK's install to be able to work with the new stuff. [12:53:42] "the new stuff?" [12:54:26] The new stuff we'll be writing to do auto-install of dependencies. [12:54:32] ah CPAN [12:54:34] PPAN [12:54:39] Not really. [12:54:49] There are several steps to go above what the sprint plan is. [12:54:51] CPYPAN [12:55:24] The current goal is to make it possible for someone distributing a package to make it automatically download its dependencies if needed. [12:55:33] well, all i know is python people worry more about dependencies than perl people [12:56:17] To have a CPAN clone, there would need to be a few other things like code signing, designated download URLs, etc. [12:56:43] a central repository which isn't disputed :p [12:56:51] Yep. [12:56:52] pje, i'd like to sign up [12:56:59] * Maniac signs up hazmat [12:57:05] i've been working on a package manager, actually i'm doing it right now [12:57:10] distributed repos [12:57:14] hazmat, just give me your e-mail and I'll include you on anything I send to the other folks. [12:57:19] i'm starting to realize that i need a back tracking engine [12:57:32] But I suggest checking out the Wiki page first, just to be sure that it really relates to your goals. [12:57:38] because i have variable version depends ie. >= 1.0 [12:58:17] * redLED always wonders why don't just people use CVS for all that, and just ask CVS from sources[] to give them a certain checkpoint [12:58:45] all u need is the checkpoint's name, and all your dependencies and merging/updating can be handled externaly by CVS [12:59:09] that way you don't need to package seperately, you don't need explicit dependencies on packages, etc. [12:59:19] redLED, the goal isn't to *distribute* the dependencies with a pacakge... [12:59:32] And it's also not intended to install them if the user already has them installed. [13:00:23] i figure all that, but my claim is that the OSS model alltogether is too dynamic of packaging, anything, in the first place in most cases [13:01:04] why not work dynamically from the development trees of each package and keep track of nothing more than 'CVS of X tag X goes with CVS of Y tag Y' [13:02:09] if you ever *do* need to distribute a package later on, you can create it completely dynamically by asking the central repo 'make me a full build for package X including all it needs', and you get back a clean build to your specifications [13:02:55] Actually, you can do that with the NetBSD pkgsrc system. [13:03:31] And, it'll be possible to do it with the distutils thing we're making, if somebody uses ViewCVS with tarball downloads enabled. [13:06:07] CPAN doesn't do code signing [13:06:13] it's totally insecure [13:06:20] AFAICT [13:06:30] Trust The Mirror! [13:07:01] there are still md5 checksums available though, no? [13:07:14] so you could always check your content vs. authoritive repo [13:07:14] well sure, but that doesn't say much [13:07:28] the authoritive repo serves both the download and listing [13:07:34] in CPAN [13:08:06] if you can compromise the tarballs you can compromise the MD5s [13:08:12] it just needs a signed index [13:08:29] but that requires additional infrastructure for verification [13:08:31] the index is generated automatically [13:08:56] ** gbay has joined us [13:09:05] so again, if the tarball is poisoned then the index would have an incorrect md5 sum AND be signed [13:09:16] you could require a gnupg key upon initial submission and not accept any further submissions unless they are signed by the key [13:09:18] incorrect as in correct for the poisoned tarball [13:09:32] meh [13:09:33] yes, there are solutions, but what I am saying is that CPAN doesn't do it [13:09:42] and we can't put crypto in mainline Python [13:09:47] CPAN was built in an era of trust [13:09:52] even rotor is coming out of Python [13:09:54] and still works [13:09:59] rotor? [13:10:01] why so? [13:10:07] U.S. law? [13:10:15] it's not US laws [13:10:17] it just laws in general [13:10:26] programmers aren't lawyers [13:10:40] so the core python people decided that they didn't want to deal with import/export laws for N countries [13:11:09] well from the module documentation: [13:11:15] Deprecated since release 2.3. The encryption algorithm is insecure. [13:11:20] yes [13:11:23] but it will actually be removed [13:11:26] at some point [13:11:40] etrepum, make a wise weblog entry. i am bored [13:12:09] I haven't had PyDS running in days.. :) [13:12:31] everything I've been working on this week is highly Mac specific [13:12:31] does any of you folks has anything good to say about Zope? [13:13:05] redLED, depends what you need it for. we've all used it i think [13:13:29] etrepum, mac bigot [13:15:05] Zope X3 has a pretty nice publishing component. [13:15:22] peak.web and the PEAK DDT tools use it to serve pages. [13:15:25] i need some generic way of doing UI, normally i'd pick wxWindows, but wxWindows, allthough cross-platform still doesen't seem to have an HTML frontend (would be difficult to do i have to agree) [13:15:52] so i'm thinking Zope for the web-based UI framework, and wxWindows for hard UI framework [13:16:06] HTML frontend for a GUI framework at that level would be *obscene* [13:16:07] wxPython ontop of wxWindows, even [13:16:51] redLED, i'm not sure zope is what you want unless you swallow the whole pill [13:17:14] it doesn't do great (Z2) at reusing existing components of your GUI app [13:17:33] Zope 3 will change all that (someday, maybe) [13:18:56] ....still bored [13:19:05] go code or something, christ :) [13:19:11] * redLED points Maniac to http://www.happytreefriends.com/ [13:20:10] etrepum, i'm trying to decide what to work on. i didn't sleep much last night so i have a severe lack of motivation [13:20:25] i'm @ work and don't feel like working [13:20:37] my real job is not coding but it could pass the afternood :) [13:20:45] * Maniac looks at the paperwork [13:36:08] * redLED looks at a bunch of apache mod_rewrite statements [13:56:40] * Maniac looks out the window [14:11:50] the real world... [14:11:53] beautiful, ain't it. [14:14:41] redLED: it's an illusion, you know that don't you? [14:33:12] i have suspected [14:42:11] i see snow [14:45:19] I see dead people, or wait.. no I don't [14:50:44] see this is what irc is all about. mindless dribble to pass the time with a little technical talk in between [14:54:21] * pje is off to a meeting. [14:54:24] Laters. [14:54:32] * redLED nods at Maniac [14:54:34] ** pje has left IRC ("Client exiting") [14:54:37] i don't know how long you've been here [14:54:46] but there are IRCnet works like IRCnet [14:54:49] with 120,000 users [14:54:53] and some 30,000 channels [14:55:06] 90% of it is mindless dribble in between mindless dribble :D [14:55:53] the other 10% are channel mode changes :D [15:00:48] ** gbay has left IRC ("It's not like I'm leaving or anything...") [15:07:58] freenode has better signal to noise ratio [15:09:22] freenode is allright besides #linux and #gentoo and #debian and such which attract too much 'customer support' [15:09:22] then again, i guess that's what they are there for. [15:53:31] ** gpciceri has joined us [15:58:57] ** gpciceri has left IRC ("Client exiting") [18:09:13] ** redLED has left IRC (Client Quit) [21:24:57] ** hazmat has left IRC (Connection timed out) [21:25:36] ** hazmat has joined us [23:38:27] ** _Maniac has joined us