**** BEGIN LOGGING AT Mon Mar 21 23:59:56 2005 Mar 22 00:06:45 perlguru-work, thanks for the update. Mar 22 00:10:29 np Mar 22 00:12:50 Slugbug 18 may be resolved in the future for unslung. Mar 22 00:13:14 then we should reopen it then? Mar 22 00:13:21 or just leave it hanging? Mar 22 00:13:28 I would leave it alone for now. Mar 22 00:13:38 its going to take significant work. Mar 22 00:14:03 Kergoth patched the hell out of a bunch of stuff to that end; but as of today, it still will not work. Mar 22 00:14:20 okay so that one could be hanging for quite a while. Mar 22 00:14:36 The reply posted should suffice for now.. Mar 22 00:14:44 okay Mar 22 00:16:15 you don't report bugs...you fix emn! Mar 22 00:16:58 Basically yes. Mar 22 00:17:02 :) Mar 22 00:17:20 catch you tommorrow...I'm beat...flat forehead today. Mar 22 00:22:41 03ingeba * 10unslung/make/findutils.mk: Added man pages and moved control files into make file Mar 22 00:24:38 03ingeba * 10unslung/sources/findutils/ (control control-doc): Moved to makefile Mar 22 02:14:59 #18 should stay open. It's still a goal if someone can work out how to do it. Mar 22 02:17:20 OK, anyone built Unslung 4.5-alpha yet? Mar 22 02:17:28 Are we devoid of alpha testers? Mar 22 02:18:43 <[g2]> rwhitby, it appears in Openslug that dropbear does require a password for root Mar 22 02:18:53 <[g2]> are you seeing the same thing with 4.5 ? Mar 22 02:19:06 <[g2]> and good morning sunshine :) Mar 22 02:19:13 its not a problem with unslung I dont think, because the rootpassword is set. Mar 22 02:19:20 (and we dont know it) Mar 22 02:19:23 Unslung hasn't moved up to the latest dropbear yet Mar 22 02:19:39 <[g2]> nod. Mar 22 02:19:57 did our DISTRO_TYPE setting not fix that? Mar 22 02:19:57 ugh, I mean "And we know it" Mar 22 02:19:59 <[g2]> this week I'm gonna figure out how to lock all the packages down in OE for Openslug Mar 22 02:20:15 I saw some interesting bk commands for that. Mar 22 02:20:28 <[g2]> bk tag ? :) Mar 22 02:20:40 bk tag won't do it for you. Mar 22 02:20:47 that would freeze *everything* Mar 22 02:20:53 <[g2]> I was teasing :) Mar 22 02:20:56 <[g2]> nod. Mar 22 02:21:21 you will want to set PREFERRED_VERSION for all the packages you want to freeze. I think there is a bitbake or OE class for that. Mar 22 02:22:32 and a problem will occur when the #oe guys just move versions instead of creating a new .bb file for a new version. Mar 22 02:22:47 So I'm not sure how feasible it will be in the long term. Mar 22 02:23:21 <[g2]> maybe a little .inc that includes all the preferred versions that ppl could include in their local.conf Mar 22 02:23:46 yeah, there's an example of that already for another distro Mar 22 02:24:02 <[g2]> I think ppl are way tired of not knowing what's going to work and what isn't Mar 22 02:24:04 but the problem still remains that the #oe guys bk move instead of bk copying Mar 22 02:24:15 so your preferred version may disappear Mar 22 02:24:43 <[g2]> ok then bk export a tree as a tarball and snapshot all the sources Mar 22 02:25:22 that's basically the same as bk tag. Which is not a problem. Mar 22 02:25:34 <[g2]> that's what I was going to do for beta Mar 22 02:25:50 I got the impression that you wanted to freeze some packages, but accept changes for others, whilst leading up to a beta. Mar 22 02:26:07 I built unslung-4.5. Havnt had a chance to fool with it yet. Mar 22 02:26:37 fully agree on the snapshot for the beta. bk tag or bk export for that. preferred versions is not sufficient cause the repo will change under you and invalidate the version you want. Mar 22 02:26:46 (as we saw for base-password) Mar 22 02:27:17 <[g2]> well maybe we stop pulling from OE Mar 22 02:27:27 dyoung <- always the trooper Mar 22 02:27:39 <[g2]> users can bk pull http://oe-devel.... Mar 22 02:28:43 I was intent on killing all the endian issues in spca5xx, but I think its not going to happen until I more fully understand the USB Specific endian issues. Mar 22 02:29:38 [g2], are you suggesting that we have Yet Another Repo for openslug-beta-release-20040320 and so on ? Mar 22 02:29:43 <[g2]> I dunno what the answer is yet for the repo thing Mar 22 02:30:17 I suggest we just live with the dynamism of the repo, and snapshot for the beta. Mar 22 02:30:40 if people want a stable repo, then just don't pull. it's as simple as that. Mar 22 02:30:44 <[g2]> I think there is a middle ground somewhere between everybody being a full-on OE-dev and a plain user Mar 22 02:30:51 tar file. Mar 22 02:31:05 yep Mar 22 02:31:50 openslug-beta-20050320-bbfiles.tar.gz or similar. Mar 22 02:32:06 or oefiles if you want to be proper. Mar 22 02:32:25 03rwhitby * 10unslung/Makefile: Added muxsshssl to CROSS_PACKAGES_READY_FOR_TESTING. Mar 22 02:32:57 I forgot to thank Kergoth for his help last night, I never would have guessed that problem. Mar 22 02:46:13 So I think that for Unslung we should make more use of the xinetd package. Anything that runs from inetd should depend upon xinetd, and just install a .conf file in /opt/etc/xinetd.d/... Mar 22 02:46:23 okay Mar 22 02:46:28 (e.g. vsftpd, atftp, etc) Mar 22 02:46:34 i nominate myself for st00pid of the year Mar 22 02:46:44 seconded :-) Mar 22 02:46:47 heh Mar 22 02:46:48 (just kidding) Mar 22 02:47:00 i just plugged in a 110v japanese bluray recorder into 220v mains Mar 22 02:47:08 nice sparks Mar 22 02:48:04 BZZT Mar 22 02:48:29 it's smelly... the power supply didn't like it Mar 22 02:48:41 well, i hope the guys at our repair lab can fix this Mar 22 02:51:16 last time I plugged 110 doodad into 220 mains and something blew up, it was just a big electrolytic cap that cooked. Mar 22 02:51:40 i hope it's something like that Mar 22 02:51:48 after all, it is a device in the $5k range Mar 22 02:52:10 ow. Mar 22 02:52:20 Same with the one I plugged intoo. Mar 22 02:52:32 well, we weren't really using it anyway Mar 22 02:52:41 It still sucks to stink up the room. Mar 22 02:53:43 true Mar 22 02:55:48 hope they can put the magic smoke back in .... Mar 22 02:56:43 or else we won't be able to record blu-ray discs with NTSC content Mar 22 02:56:49 big deal :) Mar 22 02:59:23 You could send it to me. ;-) Mar 22 02:59:38 I'm sure I can find *some* use. ;-) Mar 22 02:59:45 what? the device or the ntsc content? Mar 22 02:59:58 The Device of course! Mar 22 03:00:02 heh Mar 22 03:00:12 I have enough ntsc content floating around. ;-) Mar 22 03:00:22 ... floating through the air... Mar 22 03:00:26 ... Like Magic.... Mar 22 03:00:37 http://www.sony.jp/products/Consumer/BD/products/BDZ-S77/index.html Mar 22 03:00:42 it's this thing Mar 22 03:01:58 neat Mar 22 03:02:43 Seems to have a lot of stuff in it. Mar 22 03:03:39 ok, lunchtime Mar 22 03:03:57 23G. Wow. Mar 22 03:04:01 I want one. ;-) Mar 22 03:11:49 * [g2] needs to find a minipci ide/sata/raid/usb card other than the NEC chip which is available Mar 22 03:12:13 <[g2]> the GigE card is working quite well Mar 22 03:12:15 comfort Mar 22 03:12:35 i've seen a IDE someplace. Mar 22 03:12:43 whered I see that nowe... Mar 22 03:12:46 <[g2]> I know you guy are *there* for me Mar 22 03:13:33 "File not found" Mar 22 03:13:43 Ask me again tommrow when the brain is sorta working. Mar 22 04:01:05 bob_tm-away: ping Mar 22 04:01:33 bob_tm-away: I've updated xinetd to default to /opt/etc/xinetd.conf (so you don't have to use the -f switch). Mar 22 04:01:44 (patch to xconfig.h) Mar 22 04:01:57 rwhitby: thanks. Mar 22 04:01:59 also modified the .mk file to generate the control file (following your lead) Mar 22 04:02:14 rwhitby: Following Josh's lead, actually. Mar 22 04:02:31 Credit where credit is due Mar 22 04:02:32 yep, but you've started updating old packages too :-) Mar 22 04:02:57 Yeah - once I'm mod'ing them anyway, I do that too. Mar 22 04:03:22 Got "bb unslung-image" to work today. Mar 22 04:03:34 yeah, full credit to Josh for spearheading generating control files Mar 22 04:03:41 cool. Mar 22 04:03:42 Did a new pull - have made a new .config Mar 22 04:04:06 Will check that modules are made ok, but will only test the image Mar 22 04:04:10 Can you think of a way that we can safely automatically start xinetd without the user having to manually copy the rc.xinetd file? Mar 22 04:04:36 Yes - this is a geenral thing. I would like an easy meny thingy to put in postinst Mar 22 04:04:43 s/meny/menu/ Mar 22 04:04:58 I'd like for things like vsftpd to depend on xinetd and for it all to happen automatically Mar 22 04:05:12 then packages just drop a foo.conf file in /opt/etc/xinetd.d Mar 22 04:05:19 and restart xinetd Mar 22 04:05:20 Like "Do you want to activate xinetd now?" - "Linksys inetd is currently active - do you want to deactivate it?" Mar 22 04:05:38 I think that the act of installing a package is implicit agreement Mar 22 04:05:59 Well, if informed, it is. Usually it is not that informed. Mar 22 04:06:27 Sometimes killing a process is sawing of the branch you're sitting on. Mar 22 04:06:32 I guess the only danger is that telnet doesn't start, right? Mar 22 04:06:52 right, maybe restarting xinetd is too risky Mar 22 04:07:02 but a kill -HUP should reread the conf ... Mar 22 04:07:07 ... or that something is running. Also, I would like to do migration. Mar 22 04:07:19 Automigration of usual inetd entries Mar 22 04:07:23 kill -HUP should be safe enough Mar 22 04:07:48 It is, but it depends on the service at hand. Mar 22 04:08:18 if all we are doing is dropping in a new conf, and kill -HUP, then that should be ok Mar 22 04:08:20 I don't want to HUP a process which is doing Gods work at the time, like my ftpd Mar 22 04:08:31 Yes - dropping in new configs are ok Mar 22 04:08:41 kill -HUP of xinetd doesn't affect running service Mar 22 04:08:47 Nope Mar 22 04:08:51 (as far as I know) Mar 22 04:09:47 I was thinking of killing other services to replace the newly installed. I always like to ask the user. If some clever shell person could make idiot proof easy menus like the IPKG "overwrite config" meny Mar 22 04:10:20 ... I would add it to my packages. Proftpd, syslog-ng, xinetd etc. Mar 22 04:10:29 * rwhitby hides Mar 22 04:10:34 :-) Mar 22 04:11:55 Another thing - Ipsec support. I had the wild idea of testing it now, but I think we have to postpone it. Mar 22 04:12:08 yeah 5.x :-) Mar 22 04:12:23 I have PPP over SSH running. Mar 22 04:12:32 It is slow, but good enough for now. Mar 22 04:12:43 I will make a wiki page on it Mar 22 04:13:01 OpenVPN runs, but is not stable Mar 22 04:13:04 ok, I'm going to push the xinetd changes now Mar 22 04:13:10 np Mar 22 04:14:09 03rwhitby * 10unslung/ (4 files in 2 dirs): Now defaults to /opt/etc/xinetd.conf, and generates the control file. Mar 22 04:14:11 no change to startup procedure :-) Mar 22 04:15:26 rwhitby: btw - I think I saw some major changes to .config, like Appletalk? Mar 22 04:15:32 compiled in Mar 22 04:16:14 yeah, it had been tested by ka6sox in -able. I thought it was a module. Mar 22 04:17:01 I thought so too, but nope Mar 22 04:17:15 not in the current unslung-kernel Mar 22 04:17:23 CONFIG_DEV_APPLETALK=y Mar 22 04:19:05 apple...talk Mar 22 04:19:08 that's like Mar 22 04:19:10 soooo 1989 Mar 22 04:19:36 bob_tm-away: looks like CONFIG_ATALK should be =m. Mar 22 04:19:53 DEV_APPLETALK=y is correct Mar 22 04:19:54 I had that in my config, so it suprised me. Mar 22 04:20:05 Ok Mar 22 04:20:28 So I'll change CONFIG_ATALK in my test build here Mar 22 04:20:38 yep, and I'll update the repo Mar 22 04:20:46 Oh dear Mar 22 04:20:50 I will try. Mar 22 04:21:11 Have to contact Siddy ASAP - seem like NET_SCHED is broken Mar 22 04:24:35 03rwhitby 07 * r1.3143.1.1 10openembedded/packages/linux/unslung-kernel-2.3r25/defconfig: Changed CONFIG_ATALK from =y to =m in Unslung kernel Mar 22 04:26:52 03rwhitby 07 * r1.3146 10openembedded/packages/linux/unslung-kernel_2.3r25.bb: Bumped the Openslug kernel PR Mar 22 04:40:16 rwhitby: Is there an easy way for me to get the CVS names, full names and email-addresses of all registered developers to put into slugbug? Mar 22 04:40:45 CVS name @ users.sourceforge.net Mar 22 04:40:56 look at the Developer's page on sf.net Mar 22 04:41:15 http://sourceforge.net/project/memberlist.php?group_id=116564 Mar 22 04:42:30 Hi, how can I force my nslu to connect with 100 Mbit Full Duplex ? Are there other ways to improve the performance (now 600kb/s) Mar 22 04:46:43 rwhitby: thanks Mar 22 04:50:30 Seems that no one (except me) have a performance problem with the nslu2 ?! Mar 22 04:52:42 head12: have you checked the mailing list archives? Mar 22 04:53:53 head12: this is a development discussion channel, not an end-user support channel (that's #nslu2-general), so I wouldn't expect anyone here to have performance problems .... Mar 22 04:54:06 my suggestion would be to ask on the nslu2-general mailing list Mar 22 04:54:19 (after checking the archives, of course) Mar 22 05:01:34 wow, that's a whole lot more elaborate than the usual ~nslu2-general :) Mar 22 05:11:12 thx. Mar 22 05:26:50 OK, an interesting side effect of loading the rootfs from /dev/sda1 in unslung 4.5 is the following: Mar 22 05:26:56 # ls -ld / Mar 22 05:26:57 drwxrwxr-x 25 admin everyone 4096 Mar 22 12:04 / Mar 22 05:30:51 hum Mar 22 05:33:01 that's cause /dev/sda1 is also mounted as /share/hdd/data, and rc.bootbin chowns and chmods that for me ... Mar 22 05:38:43 night all Mar 22 06:55:18 rwhitby-asleep: Checked - current Unslung 4 alpha has same /tmp problem (not sticky nor world writable) Mar 22 07:19:39 omfg Mar 22 07:19:44 gas prices up again Mar 22 07:20:32 giel: It's us, the oil producing nations, that are sucking your wallets dry. Mar 22 07:21:01 it is Mar 22 07:21:07 but it also is our government Mar 22 07:21:10 Wish I could benefit Mar 22 07:21:24 i think we've got like the highest gas price worldwide, or something like that Mar 22 07:21:34 Which is? Mar 22 07:21:48 for 98 octane, it's 1.34 euros/liter Mar 22 07:21:54 that is $6.78/gallon Mar 22 07:21:56 Yes, that is more than here Mar 22 07:22:04 We have €1.25 Mar 22 07:22:13 Yesterday, that is Mar 22 07:22:18 the difference is pbly extra tax over here Mar 22 07:22:28 Yeah. Mar 22 07:22:34 Oh, 98? Mar 22 07:22:49 Sorry . €1.25 is for low octane Mar 22 07:23:04 I would guess about the same for 98 Mar 22 07:23:13 ...same as your price, that is Mar 22 07:24:17 low octane is 1.295 Mar 22 07:24:30 difference is less than i'd expected Mar 22 07:24:48 Then again, Norwegians tend to earn way too much money, so it doesn't hurt that much Mar 22 07:25:12 agreed Mar 22 07:25:20 it's not like i can't afford it ;) Mar 22 07:25:32 Nice to have something to complain about Mar 22 07:25:38 always Mar 22 07:25:44 Especially things like that. Mar 22 07:25:58 not keeping my engine in the high 5000 rpms would also help :) Mar 22 07:26:29 I don't even have a car anymore. Then again, I don't have income either, so just as well. Mar 22 07:38:27 03bzhou * 10unslung/make/snownews.mk: added gettext to SNOWNEWS_DEPENDS Mar 22 08:48:39 rwhitby-asleep: Doing testing with traffic-shaping on Unsling 4.x - looks pretty good so far. Siddy is a hero! Mar 22 08:49:25 <[g2]> bob_tm-away, is that with 1 or 2 interfaces ? Mar 22 08:49:30 One Mar 22 08:49:45 <[g2]> what kind of shaping ? Mar 22 08:49:49 Only ixp0 Mar 22 08:50:24 Limiting outgoing traffic to the Internet (leaving local access). Min and max thresholds based on protocols and ports Mar 22 08:50:35 Priorities among them as well Mar 22 08:50:39 <[g2]> nod. Mar 22 08:50:53 <[g2]> so the web/torrent server doesn't swamp your net connection Mar 22 08:50:56 So priorities SSH > FTP > Torrent. Mar 22 08:51:15 <[g2]> web/ftp/torrent :) Mar 22 08:51:27 Yes. especially outgoing on ADSL, wihch is usually quite lame (I only have 512kbps) Mar 22 08:51:51 <[g2]> netfilter ? Mar 22 08:52:04 <[g2]> netfilter + shaper ? Mar 22 08:52:06 Nah, based in iproute2 and iptables I believe Mar 22 08:52:13 Shaper Mar 22 08:52:23 <[g2]> iproute2 + Shaper Mar 22 08:52:24 Look on the wiki - I use Siddy's setup Mar 22 08:52:28 Yep Mar 22 08:52:37 (With minor mods) Mar 22 08:52:39 <[g2]> hey the wiki what a novel idea :) Mar 22 08:52:46 :-D Mar 22 08:53:09 <[g2]> got the url handy ? Mar 22 08:53:35 It is so neat - I can do hard FTP from the Internet while streaming video to my xbox. Mar 22 08:53:38 <[g2]> http://www.nslu2-linux.org/wiki/HowTo/EnableTrafficShaping Mar 22 08:53:41 Yep Mar 22 08:54:10 (well I could always do that, but know my IP phone works too) Mar 22 08:54:20 <[g2]> :) Mar 22 08:54:44 I guess I have to bite the bullet anyway and go to G.729 . Mar 22 08:57:15 Perfect balancing on FTP - not getting fooled by RTT or anything. Mar 22 08:57:59 * bob_tm-away is violently happy about a silly little thing and will go make a cup espresso to celebrate. Mar 22 08:59:11 No CPU use of any significance either. Mar 22 10:33:15 a quick question for the mighty kergoth: :) Mar 22 10:33:52 Would running my Computer vision process/thread with SCHED_RR and max priority give it the absolute most processing time? Mar 22 10:37:06 running it on the bare hardware without a kernel and with all interrupts disabled would give it the most processing time :P Mar 22 10:37:17 yeah. Mar 22 10:37:18 those are both just hints to linux's scheduler that your process is more important than the others Mar 22 10:37:26 But that would be counter-productive. :) Mar 22 10:37:40 Well, what do you suggest for a robot controller Mar 22 10:38:00 I have a Computer vision process that requires a lot of processing power. Mar 22 10:38:34 well, fiddling with priorities won't really give you any more processing power- there are still the same number of tasks the kernel has to find a way to execute Mar 22 10:38:43 On my slow p3-900 portable the algorithm worked fine and under 175ms but on the slug it's more like 1.2seconds! Mar 22 10:38:43 it will just improve your latency probably Mar 22 10:39:01 I'd suggest optimizing your algorithms for ARM Mar 22 10:39:13 yeah, or I could remove as many processes/daemon as possible. Mar 22 10:39:22 that's going to give you tiny improvements at best Mar 22 10:39:24 I will try the CONFIG_PREEMPT thing too. Mar 22 10:39:28 no Mar 22 10:39:33 forget about schedulers and stuff Mar 22 10:39:35 profile your code Mar 22 10:39:40 see where it's actually spending the most CPU Mar 22 10:39:42 then optimize that Mar 22 10:39:42 I did that too of course. Mar 22 10:39:55 tweaking the scheduler isnt going to help that Mar 22 10:40:00 as scanline already said Mar 22 10:40:03 when you say optimize for ARM. What do you mean specifically? Mar 22 10:40:04 theres only so much cpu time to go around Mar 22 10:40:10 nod Mar 22 10:40:25 VoodooZ_work: no floating point, certain operations are really fast on ARM (bit shifts), maybe use assembly Mar 22 10:40:32 ok. Mar 22 10:40:42 most importantly, no floating point Mar 22 10:40:47 learn to love fixed point Mar 22 10:40:51 it's really speedy on ARM Mar 22 10:41:07 I noticed that the slowest part was the erosion/dilation cv operations that scan every single pixels of a frame and compare it with a small structured element (array) Mar 22 10:41:30 Perhaps the slug doesn't like random memory access. Mar 22 10:41:42 are you using OpenCV, or something home-rolled? Mar 22 10:41:48 I also downsampled my 352x288 image to 176x144 which cut the time in 4. Mar 22 10:41:55 home rolled Mar 22 10:41:58 right, that's to be expected Mar 22 10:42:00 definitely more fun. Mar 22 10:42:02 heh Mar 22 10:42:04 and faster too! Mar 22 10:42:21 nothing too complicated though as it's only a 266mhz arm after all. Mar 22 10:42:31 so if you're already avoiding floating point and it's about as fast as you think it can get in C, you might try something like intel IPP Mar 22 10:42:38 or learn ARM assembly Mar 22 10:42:58 IPP is commercial though and I thought it was for x86? Mar 22 10:43:14 hmm, I thought they had an ARM version, but I could be remembering wrong Mar 22 10:43:23 and I think it's one of those free-for-noncommercial deals Mar 22 10:43:26 you could be right too. Mar 22 10:43:35 cool. I'll look it up. Mar 22 10:43:51 OpenCV already supports it I think Mar 22 10:43:54 the Xscale doesn't do MMX though right? Mar 22 10:43:57 no Mar 22 10:44:21 MMX is just an x86 instruction set extension, ARM has no need for such things Mar 22 10:44:53 yep. Mar 22 10:45:19 I guess 200ms is not too bad to start. Mar 22 10:46:05 It won't leave much for my AI code though but for this application it's not a big deal (Nerf-bots) Mar 22 10:46:24 hehe, fun Mar 22 10:46:36 yep. check it out at www.ottawarobotics.org Mar 22 10:47:08 it used to be Splat-bots (real paintball guns) but some people deemed it too dangerous. wimps! Mar 22 10:48:08 So I'll be using an omni directional mirror to locate ennemies and track enemies using my turret. Mar 22 10:48:22 5fps might be a bit slow though. we'll see. All for fun. Mar 22 10:48:40 hehe Mar 22 10:49:02 so you point the camera straight up at a convex mirror? Mar 22 10:49:08 yep. Mar 22 10:49:10 cool Mar 22 10:49:20 The problem is finding an appropriate mirror for cheap though Mar 22 10:49:37 I tried a christmas silver ball but it distorts object far away. Mar 22 10:49:38 yeah. a ball bearing like they use for light probe photography would be great, but really heavy Mar 22 10:49:55 light probe photography? Mar 22 10:50:26 photographing a small mirrored ball at various exposures to get a complete high dynamic range environment map Mar 22 10:50:33 I'm open to suggestions as I haven't found the perfect mirror quite yet. Mar 22 10:50:35 lets you sample a real lighting environment and use it for computer generated scenes Mar 22 10:50:53 cool. Mar 22 10:50:53 Any idea where I could get one of those? Mar 22 10:51:25 I could rummage for some links I found a while ago, but they usually use ball bearings, which would be awfully heavy for a small robot Mar 22 10:51:50 yeah, but I'm willing to try. Mar 22 10:52:09 aha Mar 22 10:52:10 http://www.ict.usc.edu/graphics/HDRShop/main-pages/faq.html Mar 22 10:52:16 I heard of silver coated light bulbs too but couldn't find any. Mar 22 10:52:21 under "Where can one buy mirrored spheres" Mar 22 10:52:22 thanks. I'll look it up. Mar 22 10:52:48 btw, there does appear to be an ARM IPP free lib. Mar 22 10:52:55 neat Mar 22 10:53:35 I'll see how practical it would be though as it might be counter-productive if the opencv code is bulky. Mar 22 10:53:46 opencv is indeed bulky Mar 22 10:53:57 And I'm sure it is as it's made to be more flexible as opposed to mine which is app specific. Mar 22 10:54:03 right Mar 22 10:54:17 But perhaps I can "borrow" some of their assembly code. Mar 22 10:54:26 might be tough to learn though. Mar 22 10:54:35 I don't think opencv has any ARM assembly, it just has an interface to IPP Mar 22 10:54:42 so you might as well just use IPP directly in your inner loops Mar 22 10:55:04 ummm. I guess I could do that. Mar 22 10:55:17 Thanks for the suggestions. Mar 22 10:55:21 sure Mar 22 10:56:01 If you get any great ideas for mirrors let me know. Mar 22 10:57:38 The link you gave me looks great. Mar 22 11:01:56 I guess I should try playing with gprof too to gain a better understanding of where the real bottleneck is. Mar 22 11:34:55 03bzhou * 10unslung/ (make/snownews.mk sources/snownews/configure.patch): changed from mkdir to install -d **** BEGIN LOGGING AT Tue Mar 22 14:36:25 2005 Mar 22 18:10:40 Does anybody know what the default root password is for openslug? Mar 22 18:12:21 <[g2]-away> novacane_22, there is none right now Mar 22 18:12:57 yeah, I tried that but it does not like it Mar 22 20:01:21 hello rwhitby-web Mar 22 20:01:36 g'day Mar 22 20:01:43 <[g2]> hey mate Mar 22 20:02:20 rwhitby-web, did you get that forwarded email? Mar 22 20:02:25 yep Mar 22 20:02:36 [g2], see pm Mar 22 20:02:53 so we need someone to write some PHP, and we won't be able to use the SF file release area any more ... Mar 22 20:03:16 should we do this in #openslug? Mar 22 20:03:23 this conversation, that is? Mar 22 20:06:00 <[g2]> rwhitby-web, what's up with the PHP and SF file release ? Mar 22 20:06:25 for a click-through download Mar 22 20:06:34 we probably should do it for unslung too Mar 22 20:06:46 for the openslug binary Mar 22 20:06:58 yeah Mar 22 20:07:05 and the unslung binary Mar 22 20:07:44 that's what I read in that email... "ok with click through" Mar 22 20:08:00 and maybe even put the license in the openslug-feeds.ipk postinst script which gives access to the ixp ipks. Mar 22 20:08:16 I'm putting it together tonight. Mar 22 20:08:22 the "clickthru" Mar 22 20:08:30 modifying the Snapgear one. Mar 22 20:08:37 ah... Mar 22 20:08:45 I was about to ask about the license.. Mar 22 20:09:05 <[g2]> I don't know how the Snapgear applies to Openslug Mar 22 20:09:11 same code Mar 22 20:09:12 <[g2]> we're 2.6 based Mar 22 20:09:46 its under the same license. Mar 22 20:11:26 so snapgear has a click-through intel license that we can use for our click-through license for openslug and unslung ? Mar 22 20:11:33 sort of Mar 22 20:12:01 they have a click through questionairre when you try to unzip the intel code/pathces Mar 22 20:12:29 <[g2]> we should also put the licenses in the rootfs no ? Mar 22 20:12:41 rwhitby-web, yes. Mar 22 20:13:14 [g2]: yep Mar 22 20:13:15 "do you agree to this license?" and if answered yes, then the script expands the code/patches Mar 22 20:15:32 who wants to do the "official" response to that email (from the group)? Mar 22 20:15:50 Sounds like ByronT and ka6sox have it all in hand. Do we need to do an official response? Mar 22 20:16:04 just follow directions. Mar 22 20:16:28 he did ask some questions.... to wrap it up, I would want to say "Here we are" and "here's what we've done" Mar 22 20:17:02 let's put it up, and ask him to review it. Mar 22 20:17:11 okay. Mar 22 20:17:46 sounds like a plan. :) Mar 22 20:26:44 No javascript? Mar 22 20:26:56 that is the way that Intel does it on their site. Mar 22 20:27:11 http://www.intel.com/design/network/swsup/ixp400_gnu.htm?url=/design/network/swsup/ixp400LinuxIntegrationPatch-1_2.htm Mar 22 20:28:24 (I realize this is GPL license but it seems good enough for Intel) Mar 22 20:29:34 they do the same for the normal license. Mar 22 20:29:49 so is Javascript okay? Mar 22 20:29:57 we could even use their script to do it (just give a different URL to it) Mar 22 20:30:05 I'm willing. Mar 22 20:30:20 I would rather use Intel(tm) approved code! Mar 22 20:30:36 hehe Mar 22 20:30:39 just check the license first ;-) Mar 22 20:30:56 yeah.. Mar 22 20:31:19 I'll ask ByronT to look at it with me before posting it. Mar 22 20:32:26 whee.... ;) Mar 22 21:07:43 rwhitby-web, still here? Mar 22 21:08:04 yeah Mar 22 21:08:20 rwhitby-web, see pm Mar 22 21:09:12 rwhitby: warning: no caustic liquids in mouth when listening... possible nasal passage burnage Mar 22 21:09:12 [21:07] no kiddin! Mar 22 21:10:40 ah yes, an oldie but a goodie Mar 22 21:10:48 that is great! Mar 22 21:11:21 rwhitby-web, any reason that javascript won't work on SF.net? Mar 22 21:11:40 javascript is client, isn't it? Mar 22 21:11:55 I'm not sure. Mar 22 21:12:02 I suspect so. Mar 22 21:12:13 it is Mar 22 21:40:10 03bzhou * 10unslung/ (Makefile make/tin.mk sources/tin/configure.patch): added tin to native ready for testing list Mar 22 21:41:31 03bzhou * 10unslung/make/tin.mk: specifies maintainer Mar 22 21:53:33 bzhou is productive ... Mar 22 22:00:41 before I forget, someone remind me later that I have to fix the diversion scripts in unslung so they recognise sda1 and sdb1 as valid rootfs locations Mar 22 22:01:34 :) Mar 22 22:02:22 rwhitby-away: before I forget, someone remind me later that I have to fix the diversion scripts in unslung so they recognise sda1 and sdb1 as valid rootfs locations Mar 22 22:02:51