**** BEGIN LOGGING AT Sat Jul 28 02:59:56 2007 Jul 28 03:01:05 nbd * r8204 /trunk/include/subdir.mk: allow target specific builddir list override Jul 28 08:51:43 juhosg * r8205 /trunk/target/linux/adm5120eb-2.6/config/default: [adm5120] use adm5120 flash map driver on BE targets Jul 28 09:23:42 [florian]: ping Jul 28 10:48:39 [florian]: I got it at least to boot with Agos eva-loader but it hangs here: http://pastebin.ca/636922 Jul 28 11:39:58 i'd like to maintain a full official package repository for the i386 target. is it possible to get access and some space to host it at the openwrt servers? Jul 28 11:43:07 s/i386/x86-2.6/ Jul 28 11:59:56 hello everyone - WIP-Day in the month of july has now officially found its beginning, and the next 24 hours are up for graps :) Jul 28 12:00:05 glp_home? Jul 28 12:01:30 I made a small list of subject, or questions, to discuss which was posted on the openwrt-devel mailing-list Jul 28 12:02:44 this was just suggestions, so feel free to add or ignore Jul 28 12:03:41 the 3 subjects i mentioned in the email was: Jul 28 12:04:04 1. Is there a need of a documentation-team? Jul 28 12:04:19 2. open hardware Jul 28 12:04:41 3. openwrt and copyright/ownership? Jul 28 12:09:21 and the email i'm refering to can be found @ http://lists.openwrt.org/pipermail/openwrt-devel/2007-July/001146.html Jul 28 12:26:07 well, imho the answer to question one is: obviously ;) Jul 28 12:26:16 but who's going to do it? Jul 28 12:32:43 nbd: very relevant question - maybe start with a list of candidates to lead such a team? Jul 28 12:34:22 * glp_home just taking a shower Jul 28 12:51:48 * glp_home back from shower Jul 28 13:00:40 nbd * r8206 /trunk/ (6 files in 2 dirs): more build system cleanup Jul 28 13:16:27 Question #2 - Just read the email. As for the simple node, well.. sounds like a Fon. And it's basically a reference design from the Atheros chipset. Who manufactures that Atheros has a list of authorized Atheros designers/manufacturers on their site. Jul 28 13:29:00 nbd * r8207 /trunk/ (4 files in 4 dirs): build system fixes, more cleanup Jul 28 13:30:09 Darkcobra: as much as i know it's Accton who makes the fonera2100, or am i wrong? Jul 28 13:31:04 I really don't know. I could examine my board for a mfg's marking, brb Jul 28 13:38:35 Not finding anything useful. Jul 28 13:39:28 Darkcobra: http://www.accton.com/products/product_range/21_weap/MR3201A.htm Jul 28 13:40:19 Cute. Case is a little different, wonder if it's any better ventilated :) Jul 28 13:41:42 Would there be any advantage to getting them direct from Accton, rather than buying from Fon? Jul 28 13:42:28 Is here anyone using bitlbee? I'm trying to update it to a more recent version, but it strictly ignores my CFLAGS. (it can't find openssl, so I added the path manually) Jul 28 13:43:48 Darkcobra: 1. you didn't have to hassle with fon. 2. maybe a manufacturer could be convinced to have openwrt as default firmware, with ssh enabled etc. Jul 28 13:44:00 nbd * r8208 /trunk/ (toolchain/Makefile tools/Makefile): use $(curdir) again Jul 28 13:44:27 Fon now sells them for $39.95. Used to be when they were free (or $5) Fon had incentive for people not to hack their routers. Jul 28 13:44:44 Now it sounds like they're making money on sales, or at least breaking even. Jul 28 13:45:13 Perhaps they are more friendly to the idea of selling the FON2100 for the explicit purpose of loading OpenWrt now? Jul 28 13:46:31 Just a thought, since their price from Accton is probably better than anything we could negotiate. Jul 28 13:46:45 Darkcobra: would probably break with the fon business plan, the company is selling a service Jul 28 13:48:22 Maybe so. But Linksys kinda broke with their business plan when they released the WRT54GL. Jul 28 13:48:32 ? Jul 28 13:52:09 Darkcobra: not to sure about what linksys did in that respect, you could also look at from the perspective that the company extended its market? Jul 28 13:52:52 glp: Right. I doubt the originally anticipated how many people would load their own software on their routers. Jul 28 13:52:58 *doubt they Jul 28 13:53:41 i think it was just a plot to get more money out of people that like to reflash their routers Jul 28 13:53:41 So when they switched to a less-capable model with VxWorks, they released a special version for the hackers. Like you said, extended the market. And they were able to charge a bit extra, above and beyond the cost of the additional memory. Jul 28 13:53:52 i mean they suddenly made the wrt54g really crappy Jul 28 13:54:02 and instead of fixing the mistake, the introduced the old version as a 'new product' Jul 28 13:54:14 which of course is more expensive than the old one Jul 28 13:54:38 I doubt they intended to actually make it crappy :) Just cheaper to make. Jul 28 13:54:56 ok, let me rephrase. they didn't intend to make it any good Jul 28 13:54:56 ;) Jul 28 13:55:11 Anyhoo... Fon could extend their market too. Who knows if they'll consider it, unless we ask. Jul 28 13:55:25 nbd: good rephrase! Jul 28 13:56:06 And there's the "other" fon, I can't ever remember the name, as it's an odd spelling. The one in sold in Europe. Jul 28 13:56:27 Essentially same hardware... but with a cooler switching 5v->3.3v regulator. Jul 28 13:58:39 Darkcobra: i thought it was the exact same Jul 28 14:01:00 Not quite. Wish I'd saved the site (or could remember the name).Ah yes, it's the Meraki Mini. Jul 28 14:02:18 Darkcobra: the meraki is sold by meraki, which is a different company Jul 28 14:02:49 Yes, I know. Sorry if I didn't make that clear. Jul 28 14:03:20 Darkcobra: and it's mainly sold in the US Jul 28 14:03:34 Was bringing ip up as another vendor we could approach to possibly provide an OpenWrt-friendly unit. And it runs cooler/uses less power. Jul 28 14:03:44 *it, not ip.. can't type today Jul 28 14:12:19 Darkcobra: no prob Jul 28 14:12:47 Were things this quiet at the last WIP day? :) Jul 28 14:15:13 Darkcobra: kind'o up and down ;) Jul 28 14:15:52 I think i Jul 28 14:16:04 ups typo Jul 28 14:17:42 it is all about what members turn wip-day into Jul 28 14:18:04 Well, I could describe the project I'm considering if you'd like. Jul 28 14:18:17 Darkcobra: would be really great Jul 28 14:18:58 I'm interested in OpenWrt primarily for hardware hacking. Hooking the router up to I2C, sensors, you name it. Jul 28 14:19:57 It's my understanding that methods to access the GPIO like kmod_gpio, are pretty slow, due to the overhead of switching from usermode to kernel for each individual bitbang. Jul 28 14:20:28 And then there's things like the MMC drivers, which are blazing fast. Jul 28 14:21:01 Darkcobra: i think _trine is also doing such things Jul 28 14:21:07 My idea is to produce a kernel module that would accept a simple scripting language that would describe what GPIO operations to perform. Jul 28 14:21:44 You send it your script to read a sensor, tell it when to execute, and it performs all the bitbanging and returns the data. Jul 28 14:22:36 what would be a real-world use? Jul 28 14:22:46 Would allow for much better speed, while still allowing access to any arbitrary attached hardware device. Jul 28 14:23:47 Anything that requires faster IO than things like kmod-gpio. Robotics would be a prime example, as you'd have a lot of sensors to attach. Jul 28 14:24:39 so more a generalized embedded system focus? Jul 28 14:25:25 Yep. Another advantage is if you needed to read a sensor repeatedly, at very accurate time intervals. Jul 28 14:26:04 You send the script, tell the module "execute every 10ms", and it does so. Since it's running in the kernel, it would have pretty good time accuracy. Jul 28 14:26:29 sound like a good idea Jul 28 14:26:52 that sounds rather cool Jul 28 14:27:41 As for output, you can create a similar script, send it to the module, then anytime you need to send data, just call the script with the variable data (like I2C address, and bytes to transfer). Jul 28 14:29:48 Darkcobra: don't you want to just use i2c on gpio? Jul 28 14:30:39 Well, actually I'd like to be able to hook up a lot of devices, of mixed type. I2C, I2S, SPI, etc. Jul 28 14:31:16 Even multiple serial ports. Jul 28 14:31:17 Uhh, why is glib-2 not including gmodule-2? Jul 28 14:36:18 nbd * r8209 /trunk/include/subdir.mk: add default command for dummy targets Jul 28 14:37:15 Darkcobra: so how will you begin this project? Jul 28 14:37:57 Darkcobra: are there similar kernel modules from other projects? Jul 28 14:38:57 Similar, not that I'm aware. It's either a generic and slow approach, like kmod_gpio; or a specific-device driver, like mmc. Jul 28 14:39:33 Well I'm beginning by learning Linux, for one thing. :) My first experience with it was Tuesday when I flashed OpenWrt. Jul 28 14:40:03 I have been able to compile a loadable kernel module, at least. Jul 28 14:40:28 Beyond that, the biggest challenge will be designing the scripting language. Jul 28 14:41:00 Darkcobra: from scratch (the scripting language)? Jul 28 14:41:41 Possibly. I don't know if there's anything in the open source community that can be adapted to the task. Jul 28 14:42:35 It would probably need to compile to bytecode to be fast enough for my goals. Jul 28 14:42:59 Darkcobra: it would be a good idea to make a post in the forum and ask if anybody had an idea? Jul 28 14:44:16 Sure. If there's something out there, and if it's a script language people are already familiar with, that would be a plus. Jul 28 14:44:44 Otherwise, I do have a compiler and bytecode interpreter I've written that could be adapted with some work. Jul 28 14:59:14 Darkcobra: sound like you'r moving fast ahead Jul 28 15:02:55 I'm trying. Linux is pretty strange for an old Windows dev like me. :) Jul 28 15:03:08 [florian]: Is there any reason why this ticket got accepted and all the !glib stuff got excluded? https://dev.openwrt.org/changeset/7198 Jul 28 15:03:36 i have no idea Jul 28 15:05:21 anyway the new bitlbee needs gmodule, so I'm considering writirg a patch *sigh* Jul 28 15:06:28 *gmodule2 Jul 28 15:09:04 Darkcobra: how come you'r switching? Jul 28 15:09:34 Switching? No. Jul 28 15:09:52 I had previously planned a very similar project, only on the ARM platform. Jul 28 15:10:32 But then I found out about the capability of Linux on routers.. Jul 28 15:12:06 It's like a microcontroller, but on steroids. I'm impressed. Jul 28 15:13:43 I don't think I'll be trading Windows for Linux as my desktop OS though, learning to program for Linux is just a means to this particular end for me. Jul 28 15:14:38 What does the $(1) represent (in the makefile)? Jul 28 15:14:44 Some of the things I've seen in Linux are quite nice though :) Jul 28 15:15:02 Norek: parameter 1 for the template Jul 28 15:15:38 nbd: ah, thx Jul 28 15:17:48 nbd: mind if I ask a question regarding my thread on the forum, re spontaneous reboots? Jul 28 15:30:51 Darkcobra: which one? Jul 28 15:31:38 ap mode, minstrel crashing with a rate index out out bound error as a result of communicating with an 802.11b device Jul 28 15:32:02 you asked "Did you leave the ad-hoc wifi-iface bridged to lan? Jul 28 15:32:18 were you just asking, or did you mean I should turn off bridging? Jul 28 15:33:14 i was just wondering because i didn't know if it worked or not Jul 28 15:33:32 i'm still looking into the rate control issues Jul 28 15:33:41 haven't figured out how to solve this properly yet Jul 28 15:34:07 ok, I'll try it an let you know then Jul 28 15:34:48 nbd: did you find out anything on the sta disconnect issue with bgscan on? Jul 28 15:35:55 not yet. mentor from madwifi says the code is pretty much broken. and i agree Jul 28 15:36:23 well I guess that would do it :-P Jul 28 16:37:47 nbd: [florian] doesn't seem to be around, and i'm in a hurry -- will be gone most of the day Jul 28 16:39:19 you can move the lzma patch and the two git-newsetup patches back to generic-2.6 after you apply the current http://homepage.mac.com/danielg4/rdc.patch Jul 28 17:14:32 * glp_home Having dinner :-) Jul 28 17:25:29 nbd: ^^ Jul 28 17:27:57 i think when i have time i will reorganize the patch to make lzma optional Jul 28 17:29:15 for now, it should at least be applied to rdc. the git-newsetup patches need not be optional Jul 28 17:29:32 yeah Jul 28 17:30:03 and, a working lzma doesn't hurt anything, either Jul 28 17:30:16 yeah Jul 28 17:30:44 so, to reiterate: Jul 28 17:30:47 you can move the lzma patch and the two git-newsetup patches back to generic-2.6 after you apply the current http://homepage.mac.com/danielg4/rdc.patch Jul 28 17:36:14 nbd: does the patch look ok to you? Jul 28 17:36:23 * nbd is a bit distracted right now Jul 28 19:42:00 Should probably remind that it WIP-Day is in progress Jul 28 19:42:50 I did several hours ago introduce folowing questions: Jul 28 19:43:05 Is there a need of a documentation-team? Jul 28 19:43:17 2. open hardware? Jul 28 19:43:28 3. openwrt and copyright/ownership? Jul 28 19:43:59 And I refered in this regard: @ http://lists.openwrt.org/pipermail/openwrt-devel/2007-July/001146.html Jul 28 19:47:55 question 3 is a continuation of a question about how the openwrt project should handle individual copyright versus project ownership? Jul 28 19:49:25 Does openwrt automatically takes over copyright to work which is added to the reposity? Jul 28 19:49:47 *takes/take Jul 28 19:50:03 i don't think we can automatically take over copyright Jul 28 19:50:20 so i think we'll just assume that it's not ours, but under the same license as our code Jul 28 19:52:23 is there a destinction between "ours" and "the openwrt projects" ? Jul 28 19:54:05 hmm Jul 28 19:54:09 good point Jul 28 19:54:13 glp_home: I was throwing around the idea of a release testing team for openwrt the other day Jul 28 19:54:26 ours as in belonging to the core team, so the team can do what it wants with it Jul 28 19:56:30 well, i believe this really only matters if openwrt intends to do a license change Jul 28 19:56:33 which won't happen any time soon Jul 28 20:15:24 thepeople_mobile: release testing team could be a good idea? Jul 28 20:16:31 nbd:to me it's more 'the openwrt project', not ours as in the 'core dev team'? Jul 28 20:16:47 yes Jul 28 20:19:09 nbd: there is something essential about the individual ownership, but it doesn't really become coherent before you share your work with a "project". Jul 28 20:21:16 I think there are boundaries to what the core team can do with the collected source-code? Jul 28 20:28:53 glp_home: I think it would provide a well tested release Jul 28 20:31:41 thepeople_mobile: i think it would be easy to make an announcement - have members commit to testing - and then get the release together Jul 28 20:33:25 glp_home: there should be a list of things to be tested and then signed off on for each platform Jul 28 20:34:38 ask for members who would be willing to collect the list for each platform, and then see if it works? Jul 28 20:35:31 defined list, then have it tested on all platforms Jul 28 20:35:45 bbl Jul 28 20:38:36 <_trine> there should be written procedures for each test so that everything is done uniformly Jul 28 20:39:06 <_trine> then just have a large group of 'slaves' doing the work Jul 28 20:39:36 "slaves" is probably not a good term :-) Jul 28 20:39:48 <_trine> :) Jul 28 20:40:07 <_trine> slaves has different meanings in English Jul 28 20:40:40 <_trine> it does not necessarily denote sub-servitude Jul 28 20:42:04 <_trine> one can for example talk about a slave relay or slave circuit Jul 28 20:42:25 guess where that expression comes from ;) Jul 28 20:42:41 <_trine> but I would not mind being a slave to the cause Jul 28 20:43:05 how about "drones"... in our little hive caste system :) Jul 28 20:43:14 <_trine> the root may well be in sub-servitude but the meaning has now changed Jul 28 20:43:39 <_trine> call me what you want it matters not Jul 28 20:43:52 <_trine> I will work for free Jul 28 20:43:55 <_trine> lol Jul 28 20:44:25 <_trine> but I do think written procedures are a necessity Jul 28 20:44:53 <_trine> otherwise the tests could be anything Jul 28 20:45:40 <_trine> once you have a set of written procedures testing a new build could be ultra quick and new procedures could be added as you go along Jul 28 20:47:47 well, more important to me than standard tests are tests with unusual configurations Jul 28 20:50:40 <_trine> does anyone ever use graph paper here I have just found a site that lets you print your own Jul 28 20:50:52 instead of having a fixed procedure it would probably be more important to have a group of people who are willing to do odd things, as nbd notes Jul 28 20:51:41 it seems like people are doing "odd" things all the time.. just not necessarily with the bleeding edge versions, or reporting back in when things *do* work Jul 28 20:55:33 Darkcobra: then it os needed to inspire people/members to report back Jul 28 21:27:16 * glp_home I'll be sleeping for a couple of hours :-) Jul 28 22:35:38 olli * r8210 /packages/net/unfs3/ (7 files in 3 dirs): Add unfs3 **** BEGIN LOGGING AT Sat Jul 28 23:27:25 2007 Jul 29 00:04:34 blogic * r8211 /trunk/target/linux/etrax-2.6/ (7 files in 6 dirs): some fixes to the etrax build **** ENDING LOGGING AT Sun Jul 29 02:59:56 2007