00:07:20 jmkasunich has joined #emc-devel 00:07:28 cradek: you here? 00:18:57 was curious about skunkworks results, just talked to him on #emc 00:19:04 going out tonight, won't be online 00:19:30 btw, today I used hal's freqgen and encoder modules to run a small DC servo _without_ external PWM generation 00:19:35 just an H bridge ;-) 00:19:52 neet :) 00:19:54 goodnight all 00:19:56 jmkasunich has quit 00:59:06 rayh_ has quit 01:31:18 jmkasunich has joined #emc-devel 01:38:57 evening john 01:39:51 hi ray 01:41:13 how's things in the north woods? 01:41:22 You know what apt command I use to get the sources for emc2 and all the dev dependencies. 01:41:45 Cold -15 right now but not a lot of snow in the last batch. 01:42:24 Trying to get a ubuntu ready to compile here. Got the installed version going. 01:42:38 not sure about that. 01:43:05 I did an Xubuntu install on an old laptop last night, cradek was asleep, and I ended up trial-and-erroring it 01:43:15 rayh, I think you need something like apt-get source emc2-axis 01:43:25 you may need to add in an option like --build-dep 01:43:29 unfortunately I wasn't taking notes 01:43:40 (I wasn't doing much, it was 4am when I finished) 01:44:10 I _thought_ that cradek had an easy way, I could swear he explained it when I did my original ubuntu install 01:44:16 I do remember that was pretty painless 01:44:50 jmkasunich: yup, he had 01:45:04 but I can't recall right now ;) 01:45:06 Right I did it in Milwaukee on Thursday but forgot the command also. 01:45:45 I think it's apt-get build-dep emc2 01:45:57 that sounds familiar 01:52:05 Thanks john. Got that stuff downloading. Couple of hours. 01:52:23 such fun 01:52:49 I had some fun today... 01:52:49 so, what's new? 01:52:59 * alex_joni had fun yesterday 01:53:10 we're working on small vehicle for an engineering week contest 01:53:19 two motors and two drive wheels, plus a caster 01:53:33 was in germany, and got a call from home that our main server crashed:( 01:53:38 our control is by joystick, with analog PWM generation 01:53:48 jmkasunich: that sounds nice ;) 01:53:58 how small? 01:54:00 but today I used HAL freqgen and encoder drivers, closed a servo loop with no external PWM generation 01:54:07 motors are 12V about 1A 01:54:22 about 3.5-4 cm dia, 7 cm long 01:54:28 with 64 count/rev encoders 01:54:45 we have a 45:1 worm reduction driving wheels about 10cm dia 01:55:09 Got the developer download on the way. Can sleep for a few hours while it arrives. Thanks all. 01:56:28 rayh: np 01:56:36 jmkasunich: here's a thought 01:57:09 it seems (from what I've read so far) that memory over 1G needs to be mapped temporarely into the kernel address space 01:57:15 yeah 01:57:30 and theres a HIGHMEM kernel config that gets involved too 01:57:33 to be accessible, that might be a problem for the RT stuff, which goes over the kernel 01:58:40 the thing that is interesting is that cradek was able to produce faults using only linux stuff 01:58:43 (gcc) 01:59:19 yes, but using the same rtai patched kernel 01:59:24 right 01:59:38 and you might have had something in the background doing it's own stuff 02:00:22 I believe that anything accessing memory over 1G on that box will result in memory corruption 02:00:28 no hard proof of that 02:01:09 probably so 02:01:13 am I right to assume that there is not current "testing" wiki page. 02:01:25 ? 02:01:28 though it is referenced from the release status page. 02:01:41 I know cradek tried hard to make it use >1G with the stock ubuntu kernel 02:02:11 ah sorry found it. 02:02:15 22:59 < cradek> I'm having real trouble using >1G of ram 02:02:15 22:59 < cradek> yeah I'm trying to stress it 02:02:43 ray: I don't know what you mean - the page referenced from the release status page is a reasonablely accurate description of "testing" as it stands right now 02:02:44 23:16 < cradek> skunkworks: I think we might have a rtai bug 02:02:59 23:16 < cradek> skunkworks: when I load realtime, the compile craps out badly 02:03:34 ok, thats new information for me 02:03:42 from tonight 02:04:07 Damn. The organization of "wiki" is impossible. 02:04:22 it desperately needs cleaning 02:04:25 indeed 02:04:40 btw, john wanted to ask you how you like that layout I sent you? 02:04:45 if you got it that is.. 02:05:11 linuxcnc layout? 02:05:45 yes 02:06:05 its pretty, but we have so many other things to spend time on 02:06:20 I have no problem with the style of linuxcnc and the wiki 02:06:27 the problem is with the organization 02:06:40 Does that layout expect a database? 02:06:50 it's not my time I was spending ;) 02:07:17 rayh: yes, I imagined a CMS (Content Management System), where you can administer it online 02:07:31 it allows for better organisation 02:07:31 way over my head 02:07:43 I write html with a text editor 02:07:51 (which is why I don't build many webpages) 02:07:52 jmkasunich: let me worry, you'll only have to publish articles once in a while 02:08:06 the beauty of this is that you won't have to in the future 02:08:29 just log in, and you can write like in kate (which I know you like), you have nice buttons for font, sizes, etc 02:08:46 IOW, a lot like the wiki 02:08:51 * alex_joni looks for an example if you got 2 mins 02:09:02 yes, but a bit more configurable 02:09:11 but you can say it's just a big wiki 02:09:23 although you won't have public access for changing pages 02:09:35 alex_joni: LAMP I hope. 02:09:44 that can be good or bad 02:09:48 LAMP ? 02:10:14 bad in that you don't benefit from the writing of others, good in that you don't wind up with a disorganized mess 02:10:15 Linux, Apache, MySQL, Perl 02:10:16 jmkasunich: you can set up unlimited accounts, and have them either trusted (and publish directly), or have them to be approved by someone 02:10:23 rayh: sure 02:10:39 it's joomla (www.joomla.org) I found very fitting 02:10:43 jmkasunich: try http://demo.joomla.org/administrator/ 02:11:00 just to get a feeling (don't be scared of the complexity) 02:12:28 how do I log in 02:12:34 hmm.. someone might have fumbled with their demo, not working for me right now 02:12:45 you should use Admin/Admin for user/pass 02:13:10 but I think someone might have changed the pass (it will be refreshed to a default install soon) 02:15:29 tried admin and Admin, neither works 02:15:32 must be busted 02:15:45 yes, probably so (it has public access to do anything) 02:15:55 but they have a cron to reset it every hour 02:22:14 rayh has quit 02:23:01 jmkasunich: try http://demo.gigcalendar.net/administrator 02:23:04 demo/demo 02:24:18 * jmkasunich is too tired to get involved in that tonight 02:24:29 stayed up to 4am last night 02:24:31 just 2 clicks around ;) 02:24:36 but I know how you feel 02:24:43 04:24 < alex_joni> but I know how you feel 02:24:50 ;-) 02:24:51 lol 02:25:12 I was up till midnight working with cradek on the skunkworks problem 02:25:14 I had some crazy days lately 02:25:23 yeah, seen the logs 02:25:47 I was up on wednesday on skunkworks's machine 02:25:49 shut down the box, was getting ready to go upstairs, then I realized "shit, I meant to install Linux on that laptop, and I need it tomorrow" 02:25:51 till about 2am 02:26:05 so I started the install after midnight 02:26:15 best time.. it's quiet 02:26:50 did you log in on that site? or do we do this some other time? 02:27:15 I logged in, clicked a couple times, determined it needed more brain cells than I can muster up this evening 02:27:21 ok ;) 02:27:59 about all I'm good for tonight is bullshitting 02:28:15 that's good too ;) 02:28:27 on that note: http://yarchive.net/metal/empirical_feed_rates.html 02:28:31 same state around here 02:28:32 a good read 02:28:53 from rec.crafts.metalworking some years ago 02:32:08 yeah 02:33:18 ok, I'll hit the bed 02:33:55 goodnight 02:42:34 goodnight 03:02:44 SWPadnos has quit 03:18:11 skunkworks has quit 04:12:05 rayh has joined #emc-devel 04:12:14 It's alive 04:12:28 hi all 04:12:39 hi ray 04:12:42 what's alive? 04:12:53 My compile of emc2 04:13:02 under ubuntu 04:13:11 small problem with rip 04:13:19 emcTaskInit: user-defined function /usr/share/emc/ncfiles/M101 found, but not executable, so ignoring 04:13:20 e 04:14:00 rip says that? 04:14:27 yes. It does show that it's set executable when I look at the permissions. 04:15:25 strange, I don't think rip knows that path at all 04:15:41 did you configure --enable-run-in-place? 04:15:51 Here's the set of commands I used. 04:16:18 ./configure --enable-run-in-place 04:16:26 make 04:16:33 sudo make setuid 04:17:10 let me try it here too then 04:17:23 ray's at it again (he can break anything, its his specialty!) 04:17:29 hi john 04:17:30 and to run -- rayh@rayu:~/emc2-2.0.0$ scripts/emc 04:17:58 Hi guys. I must warn. This box has both installed and rip. 04:18:06 ok, that should be fine 04:18:07 that should be ok 04:18:16 I believe I have the same here 04:18:45 Testing the picker shows that it does what it says it will. 04:19:14 now wait a second 04:19:15 I have not tried /usr/local/etc/emc2/configs but in home they work great. 04:19:23 are you picking the ini from /etc/emc2/sample-configs? 04:19:33 emcTaskInit: adding user-defined function ../../nc_files/M101 04:19:38 this is what I get with rip 04:20:08 No. That one was a copy from sample into ~/emc2/configs/m5i20 04:20:24 ok, that nc_files path is hardcoded then 04:20:35 Let me try copying from rip to home. 04:20:36 it's in your ini 04:21:10 You mean hard coded in by the installer. 04:21:26 the installer doesn't mess with the ini files 04:21:29 well yes and no, I edited those pathis in the inis that are in the deb 04:21:34 paths 04:21:42 ok, so I'm wrong 04:21:50 happens all the time ;-) 04:21:51 make install doesn't do it right now 04:22:23 the original reported problem is that M101 is not installed executable, and that is a bug 04:22:51 the hardcoded paths in the inis aren't a bug, but they may be bad design that should be fixed, I'm not sure 04:23:47 I wonder why M101 is in the nc_files directory in the first place 04:24:48 I see that the only place rip can handle configs is emc2xx/configs. 04:25:27 right 04:25:40 anywhere else, and they're not "in place" 04:25:57 I agree 04:25:59 don't want to risk stomping on somebody's real config 04:26:24 This means that if we set up a system using rip, we will need to do a bit of editing if we move it to installed. 04:26:47 yes currently you have to edit pretty much anytime you move a config. 04:27:00 it's unfortunate 04:27:09 if all files associated with a config are in the config dir, then there should be no hardcoded paths 04:27:11 I wonder if we can edit the picker to show only the one location for rip? 04:27:34 the picker is passed a list of locations 04:27:35 yes, that would go in emc.in 04:27:45 for rip, it should get only one 04:27:55 I'll fix it right now 04:28:13 what are you talking about 04:28:22 on mine, rip _does_ show only one place 04:28:25 pickconfig's selection path 04:28:34 /home/john/emcdev/emc2testing/configs 04:28:38 oh... 04:29:51 no I think if you have ~/emc2/configs/xxx or /usr/local/etc/emc2/configs/xxx you'll see them 04:30:07 maybe you don't have either 04:30:15 have as in if they exist? 04:30:18 yes 04:30:36 I mean copy a config into one of those dirs like ray did 04:30:45 I think it'll show up 04:30:50 ok, I suppose I can see how the usr/local one might get used by mistake, but why ~/emc2? 04:31:03 it's a feature 04:31:10 I can see the need for a proper config copier. 04:31:11 it's hardcoded in emc.in 04:31:20 that's what I was going to fix for rip 04:31:24 ok 04:31:36 it should be hardcoded to only one dir for rip 04:33:12 ok fixed 04:34:46 jmkasunich: the problem with skunkworks's machine seems to be: if you have more than 1G of ram AND realtime is loaded, apps (nonrealtime) crash right and left 04:35:19 jmkasunich: I borrowed some ram and hope I can put >1G in one of my trusted machines this weekend to test for that same pattern here 04:35:32 alex pointed me at a long thread somewhere explaining stuff I never new about linux and ram 04:35:43 I'm sure I don't know it either 04:35:47 in particular, linux handles ram over 1G differently 04:35:51 I know there are limits at 1G, 4G, 64G 04:35:51 IMO the easy solution is save money on RAM. 04:35:58 rayh: no kidding! 04:36:09 yea, no reason to have that much in a machine controller 04:36:12 put the extra in a winders box. 04:36:29 obtw cradek, I did a Xubuntu install on a laptop last night 04:36:33 I'd like to fix it if I can, but I won't worry too much if I can't 04:36:39 Gotta get going here. Great job on this stuff you guys. 04:36:51 goodnight ray 04:36:56 goodnight ray 04:37:12 I find a few things in gnome that are awkward compared to kde but ... 04:37:30 yeah, I'm having a hard time getting used to gnome 04:37:34 you could always try the kde version - not sure what kind of shape it's in 04:37:34 I think that we have a really good thing coming together. 04:37:39 rayh has quit 04:37:44 this is still a trial install, I might look into kubuntu 04:37:44 I'm glad to hear him say that 04:38:01 I used the xubuntu on the laptop because it is 366MHz, 128M ram 04:38:22 emc should still install fine I would guess 04:38:29 no gnome dependencies 04:38:34 well there's gtk1.2, hmm 04:38:42 the emc2-install.sh wants gksudo or something like that 04:38:51 gsudo? 04:38:57 oh yeah, just comment that out 04:39:10 it just gets the password cached for the subsequent sudo calls 04:39:11 well, my real interest was compiling 04:39:54 did you get it going? 04:39:58 at some point (I think during emc2-install.sh) the X got weird, I exited X and restart it and it was ok 04:40:13 then I spent a lot of time figuring out what packages are needed to do a build 04:40:22 apt-get build-dep emc2 04:40:24 I know theres an easier way, but I couldn't remember it 04:40:28 now you tell me 04:40:31 or man apt-get :-) 04:40:36 where were you at 3am 04:40:45 blissfully asleep 04:41:19 (I forgot that I needed to get that laptop working - I shut down my main box at midnight after we finished with skunkworks, ready for bed, then said oh shit 04:41:25 and was up till 4 04:41:33 why did you need the laptop right then? 04:41:42 I wanted to take it to work today 04:41:55 ah 04:42:11 we're building a small vehicle for an engineering week contest, and I used hal/freqgen/encoder/pid for motor control 04:42:27 neat 04:42:29 kinda cool actually, brush servos with no hardware except an H bridge 04:42:42 no univpwm or anything 04:42:54 software pwm? 04:42:57 yes 04:43:03 mode 1 of freqgen 04:43:07 how do you read the encoder? 04:43:16 software encoder HAL component 04:43:26 these are little motors from a tape library 04:43:38 1.5" dia, 2.5" long, with a 64 cpr encoder 04:43:48 I mean what's the hardware interface to the encoder? just parport? 04:43:54 yeah 04:44:00 slick 04:44:17 64 cpr means less than 10K counts/sec even at 7000 or so RPM (the motor speed at 12V) 04:44:23 and you're lucky enough to have a laptop that doesn't have realtime problems? 04:44:39 I'm not so lucky, mine is terrible 04:44:40 no problems bad enough to hurt this application 04:45:00 I'm not machining gold, I'm driving a little thing around the building lobby 04:45:05 haha 04:45:13 sounds really fun 04:45:30 I wanted to run latency, but at 4am, when it didn't work the first time I quit 04:45:42 yeah you have to mknod 04:45:52 cursed udev 04:46:03 I only fixed it for /dev/rtai_shm 04:46:12 no problem 04:46:32 one reason I wanted to take it in was to show it off 04:46:45 work uses windows exclusively 04:46:50 ah 04:47:09 we have programmers who do realtime firmware for motor drives (embedded, not doze ;-) 04:47:20 and they appreciate interesting stuff 04:47:37 I bet this was exciting to see then, knowing what's involved 04:47:50 we have an app something like halscope (but less flexible) for looking at what goes on inside a drive 04:48:01 but compared to halscope it sucks 04:48:15 you need a map file, and enter the address of the thing you want to monitor 04:48:23 ugh 04:48:41 1000 samples total, split between 2 channels (halscope has 16K, and 1/2/4/8/16 channels) 04:49:02 and the data goes over a serial link, so it takes 10 secs or more after a capture to view the results 04:49:17 halscope can capture/display several times a second 04:49:42 yeah it's definitely cool. 04:50:03 the one in the drive sucks, but its better than nothing 04:50:18 it has helped find several problems that would have been impossible otherwise 04:50:55 I don't know if I'll use the laptop/hal for the engineering week project, maybe just to play with 04:51:13 now you can hack emc at the coffee shop too. 04:51:16 the actual event involves a human controlling the vehicle, using wired or RC control 04:51:35 if I had a joystick that would fit the laptop, and a HAL driver for it, that would work 04:51:48 but as it stands, we'll probably use analog PWM and a couple pots 04:52:02 don't think I've ever seen a laptop with joystick port 04:52:14 usb joysticks 04:52:29 the driver wouldn't have to be hard RT 04:52:31 ah of course 04:52:43 user space that polls the stick 10 times a second and updates hal vars would be fine 04:53:31 anyway, after the event is over, I'm gonna unplug the analog PWM and wired controls and try to run the course autonomously 04:53:52 (blind, just doing dead reckoning) 04:53:54 yeah that's a much more interesting challenge 04:54:07 I expect that tire slippage will make it impractical 04:54:11 yeah 04:54:14 carpet? 04:54:21 shiny terrazzo 04:54:29 that's good 04:54:53 except the wheels we used (because we had them) are spiky off-road looking things 04:54:57 you should make a two-wheeler with a tilt sensor and pid loop that keeps it standing 04:54:59 4" dia, over an inch wide 04:55:15 oh yeah, that's not going to be very good I bet 04:55:57 there is a 45:1 worm between motors and wheels, so it works out to 288 encoder counts per inch (approx, I'll calibrate it by actual measurements) 04:56:11 wheels are about 12 inches apart 04:56:36 so without slip I could control heading to a moderate fraction of a degree 04:57:00 sounds like you need rubber, not spikes 04:57:08 but slip and ambiguity about wheel spacing (because they're wide) will make it much worse 04:57:33 yeah 04:57:41 I want a big honking O-ring 04:57:56 1/4" thick, 4" diameter 04:58:10 would probably fit nice and tight on the wheels 04:58:13 yeah 04:58:21 low slip, and narrow contact patch 04:58:23 you could stretch it around the wheels you have? 04:58:29 probably 04:58:47 they are rubber tires, with spiky and blocky rubber treads 04:58:51 I just went to the hardware store to buy an o-ring tonight 04:58:54 (from some RC toy truck) 04:58:58 they have some pretty large ones. 04:59:18 not 1/4" thick though, I think 1/8 & 3/16 04:59:37 3/16 _might_ be big enough to get out of the spikes 04:59:57 if I was serious though, I'd want to replace the rubber tires with metal with a v-groove for the o-ring 05:00:19 the tires compress a little, another source of error 05:00:23 ah 05:00:46 (this thing has to carry a 12lb bowling ball, and we're using a motorcycle battery for power - total weight will top 20 lbs) 05:01:02 motorcycle batteries are great for little robot projects 05:01:37 crap, I should have brought in my motorcycle battery 05:01:38 this one is borrowed, from a real motorcycle ;-) 05:01:44 oh well, too late now 05:01:51 ? 05:02:04 I mean it's february, too late to save the battery 05:02:25 bummer 05:02:27 I meant to bring it inside "before it got cold" 05:02:36 oh well they're not expensive 05:02:49 I just hate wasting things by being lazy 05:02:51 yeah, another team just went out and bought one 05:03:04 I like to see how cheap we can do these projects 05:03:15 yeah that can be part of the fun 05:03:17 (this is our 4th or 5th year doing eng week stuff) 05:03:33 we're recycling the H-bridges from last year 05:03:36 are you related to a college somehow, or is it just for fun at work? 05:03:45 which were made from scrap parts from the lab 05:03:48 just work 05:03:51 whenI think of e-week I think of the university 05:03:54 cool 05:04:10 we have something like 1000 people in this building, with a large percentage of engineers and other nerd types 05:04:20 I bet that's nice 05:04:36 I have 40 programmers and support people with a large percentage of ... well forget it 05:04:37 probably get 10-12 teams, ranging from those who start the night before and use popsicle sticks and duct tape, to serious competitors 05:04:48 steves_logging has joined #emc-devel 05:05:04 usually 3-4 teams are seriously in the running 05:05:07 (us included) 05:06:22 * jmkasunich wonders about re-making the wheels...... 05:06:26 what exactly is the goal this time? pilot a robot carrying a bowling ball through a maze? 05:06:34 no 05:06:35 bowling 05:06:43 (knock down pins for points) 05:06:49 ah, cool 05:06:55 the pins are in a preset location? 05:06:58 you don't have to use the ball, but you must carry it around with you 05:06:59 yes 05:07:07 different pins are worth different points 05:07:25 get to work on the lathe, it's only midnight... 05:07:29 the 10 that are 4ft in front of the balls initial position (traditional triangle layout) are only one point each 05:08:06 the one that is thru a 18" wide x 15" high tunnel and on a 16" high pedestal (8 cinderblocks) is worth 30 points 05:08:29