00:50:40 rayh has joined #emc-devel 00:51:51 Where can I get info about the ubuntu and the emc related packages? 00:52:32 for info on ubuntu in general, there's ubuntu.org 00:52:52 cradek has the info on emc packages for it 00:53:08 I'm thinking of the speciics on the emc related packages. We need a wiki page on this. 00:53:14 he has a script (from his website?) for doing the install 00:53:16 yep 00:54:22 The only ubuntu I see on the wiki is the stuff from Lawrence and Paul. 00:54:44 I'm not sure there is any more info, but I agree that there should be 00:54:49 (in the wiki) 00:55:31 I guess I'll have to visit timeguy or whatever. How you doing this evening? 00:55:39 ok, you? 00:56:06 Worked out in the wind today. Sawmill stuff. Kinda wiped out. 00:56:19 I can imagine. is it still cold there? 00:56:35 Started about 0 and got up to near 15 00:56:44 http://www.timeguy.com/cradek/emc/ubuntu 00:56:46 balmy :) 00:56:54 Thanks for the link. 00:56:58 sure 00:57:14 balmy it is. The 500mcm wire was just a bit stiff though. 00:57:28 500mcm - what's the diameter of that? 00:58:37 about 1 1/8 inch with the insulation. 00:58:48 ok - 0.850" - thick! 00:58:59 copper? 00:59:05 4 of em into a 4 inch conduit. 00:59:20 that's a bitch to bend in any weather 00:59:32 near a ton of copper in that one run. 00:59:33 even 7/8 aluminum isn't trivial 01:00:10 yep - should be about 1 pound for every 4 inches (plus the insulation) 01:00:11 A few years ago I pulled some wire like this with first time copper. 01:00:28 It was really easy to bend. Today's alloyed stuff is really a bitch. 01:00:54 heh, I'm not sure a human could do it to beryllium copper 01:01:03 and definitely not phosphor bronze 01:02:45 Right. This is just trash/copper. Recycled in these metals leaves something to be desired. 01:03:21 Anyway. The wire is in and connected at both ends. 01:03:36 cool, you have a new 600A service now ;) 01:03:47 or a very beefy DSL line 01:03:59 Now for 300 amps of 480 volt phase converter and we'll be ready to connect machines. 01:04:19 Don't I wish DSL was here, or even close. 01:04:23 heh 01:04:35 as I recall, you even have horizon troubles with satellite 01:04:53 This spring I'll put up a tower and get someone in town to host a directional 802 01:05:02 pringles cans to the rescue 01:05:09 That's it. 01:05:12 (plus you get to eat the pringles) 01:05:24 rofl 01:05:55 I thought I'd have to build a CNC parabola spinner. 01:05:59 or conversely, you have to eat the pringles (depending on diets and stuff) 01:06:13 see food diet. 01:06:14 there's a good reason to have an ellipse command in emc :) 01:06:18 me too 01:06:27 I just try to not look at crappy food 01:06:48 double bacon cheeseburger for lunch. 01:06:56 near barf all afternoon. 01:07:02 yummy - especially with mushrooms and Guacamole 01:07:30 Oh I went for the health food version, lettuce and tomatoe. 01:07:34 tomato. 01:07:46 tomatoes are fine too 01:07:54 as long as they're ripe 01:08:23 Vine ripened. It's a UP dream. 01:08:37 Happens usually one day in september. 01:09:03 yummy - especially with buffalo mozzarella, basil, and a bit of oil and vinegar (balsamic) drizzsled on 01:09:41 Yep. That is a favorite (usually sheep cheese) in Germany. 01:10:12 heh - or exactly as I described, in Italy (insalata Caprese, or Capri salad) 01:10:27 Nice. 01:10:54 also made well at Guido's italian restaurant, in Los Angeles :) 01:12:00 There are not many so called Italian restaurants in the USA that I like. Usually avoid 'em if they call themselves that. 01:12:12 yep 01:12:18 I think it's just called Guido's 01:12:21 But next time in LA I will have to visit Guido's. 01:12:28 11950 Santa Monica, I think 01:12:49 Okay. 01:13:03 oops - 11980 01:13:18 Close enough. 01:13:24 it's been a couple of years since I've been there :) 01:14:09 You thinking about attending CNC-Workshop? 01:14:15 considering it. 01:14:24 I'll have to see how the camera project goes though 01:14:48 I wonder if having a code group part of each day would be good. 01:15:13 for the workshop? I'm not sure 01:15:14 JMK and I had talked about a code week after but it makes a long time away from home. 01:15:20 yep 01:15:45 hi all 01:15:53 Hi Chris. 01:16:01 maybe run classes or retrofits half day and do coding the other half, but I'm not sure 01:16:03 hi Chris 01:16:21 I always feel like I'm missing out on whatever else is going on 01:16:28 I was thinking I could lecture on the beginning stuff while you guys code. 01:16:42 heh - "Loudmouth duties" :) 01:16:43 Then of course evenings would be taken up with code+ 01:17:10 That's it. 01:17:26 it also depends on what else is going on - there may be things that the emc folk want to attend, so that cuts into coding (or sleep) time 01:17:28 What do you think Chris? 01:18:09 My guess is that the code stuff will be as much architecting as writing. 01:18:15 Much like last year. 01:18:17 I think I'll be able to come 01:18:25 but it's my first so I don't know what it's about 01:18:30 does anyone (other than Paul) have a copy of his lathe code? 01:18:56 SWPadnos: nobody has ever seen it afaik 01:18:59 my first CNC workshop as well, if I make it 01:19:00 Would you want a code and coder only time 01:19:14 Paul said that it was not compatable with emc2 01:19:17 well - at least a time separate from the cnc workshop 01:19:21 rayh: if we have goals, I think that would be great 01:19:44 I would let you guys make up the goals for coding sessions. 01:19:45 look over the 2004 and 2005 fest goals, and I think threading will be #1 :( 01:19:57 personally I'm struggling with the concept of paying money to help people convert machines and give away (for free) the software I helped write (for free) 01:20:04 heh 01:20:10 (I'm serious) 01:20:19 yeah - I'd hope that the fest people could get in for free or something 01:20:21 I agree 01:20:43 it feels wrong since we're providers, not consumers 01:20:53 seems there should be a difference. 01:20:53 I understand. Roland does have a lot of expense. Tents and porta-potties and such. 01:21:12 We do fit into the thing a bit different than the average. 01:21:15 actually, there should be CDs available, and we should sell them for a small fee (like $5), and use the cash to fund the llinuxcnc.org website 01:21:17 I'm sure that's very true 01:21:49 well - if we can teach some classes, that shoud be taken into consideration for entrance fees 01:21:54 SWPadnos: I'd rather get in free, and give away CDs (even at my expense) 01:22:18 I plan to order ubuntu a stack of CDs 01:22:21 err 01:22:27 I plan to order a stack of ubuntu CDs 01:22:29 That would be great.\ 01:22:36 sure - I'm just thinking about getting the website onto a faster server, which costs some $ 01:22:47 not the most important thing, but still 01:22:50 true 01:22:55 You bet. 01:23:47 the dreamhost rates are really good. I can set up a discount certificate that gets $97 off the fees as well (I can just donate my referral fee) 01:23:59 s/fee/reward/ 01:24:59 dreamhost does seem good 01:25:05 I'm not convinced that we should host it on a site that has a charge for over xxx per month. 01:25:06 I think jepler has looked into them 01:25:37 rayh: I think their limits are really high 01:25:55 and the cost of on tb? 01:25:57 1.6 TB/month transfer, 60GB storage, for $15.95/month 01:25:58 rayh: but I know how you feel 01:26:12 that includes a prepayment discount 01:26:16 http://www.dreamhost.com/shared/comparison.html 01:26:30 code monster is what I have (for cncgear.com) 01:26:42 we would never get near that unless we hosted some very popular ISOs 01:26:43 I guess I just got spoiled by tbl when he was at Sherline. 01:27:00 and I doubt we'll ever host ISOs again 01:27:11 every time they have a new special, they give you the new deal (I started at 7.68G storage and 256G transfer) 01:27:12 yeah he was sure a big help 01:27:19 I don't think I'd say that. 01:27:30 An ISO is very nice. 01:27:40 For folk who are web challenged. 01:27:42 I think an ISO is pretty critical 01:27:57 if it had just our packages, it would be small, that's fine 01:28:01 I meant a 600M iso 01:28:14 it still won't get too high - cncgear is around 120G/month, and it's all the BDI ISO 01:28:16 That's exactly what I mean also. 01:28:47 well - given the choice of what I've seen of puppy, and th eBDI4.x, I'd take the 600M ISO 01:28:57 puppy is cool, but looks childish at this point 01:28:59 SWPadnos: try my ubuntu packages then 01:29:07 it's about 10 years behind the curve for UI design 01:29:18 I got another complaint today about the dead brown screen. 01:29:27 I will. I may even plop in some 64-bit ones one day 01:29:41 rayh: who? what hardware? what stage in the install? 01:29:43 I wish I had it so I could at least pretend to know what's going on. 01:29:55 I can mail you an ISO if you like 01:29:55 rayh: sure is taking a while to get your CDs to you 01:30:00 I mean, a disc 01:30:24 rayh: did it give errors during the earlier stages of the install? 01:30:26 you sent some already? 01:30:41 ubuntu will send them to you for free if you just ask. ray asked 01:30:45 but I guess they're a little slow. 01:30:47 ah - OK 01:31:09 No folk are downloading ubuntu and getting ready to try the emc install. 01:31:20 hmmm - should the ubuntu packages work for a BDI? 01:31:35 rayh: "No" is a bit of an overstatement 01:31:45 SWPadnos: don't know 01:31:54 SWPadnos: probably not 01:31:57 I'd say that no "customers" are doing that - only early adopters and developers 01:32:06 Don't think it will without some sort of apt mixed packages. 01:32:12 sure, that's because there hasn't been an announcement 01:32:20 well, they're both debian/apt based 01:32:35 SWPadnos: the depends will be wrong - kernel versions etc 01:32:37 sarge v breezy 01:32:42 right - it's better to have a single CD with the correct packages - like the BDI 01:32:44 ok 01:32:56 SWPadnos: I disagree totally 01:33:13 I should be able to --force the versions in the emc repository, but it probably isn't worth it 01:33:22 in addition to the .deb based system 01:33:40 How much trouble would it be to build a deb for BDI? 01:33:45 SWPadnos: if you run my kernel, rtai, and emc packages together, it might work 01:33:59 that's what I was thinking, but it's still too much bother ;) 01:34:01 rayh: if we picked one version and people didn't update parts of it, pretty easy 01:34:17 Okay. 01:34:18 rayh: as soon as they change kernels, rtai, etc, our package won't work anymore. 01:34:55 rayh: I am doing this because I think it's best for us to have control of all the necessary packages together. Otherwise we're always guessing how to match up. 01:35:10 Sure. I'm pretty convinced that we must think of kernel, real-time, and emc as an entity. 01:35:24 it's true (if a bit unfortunate) 01:35:37 with precompiled stuff, that's 100% true 01:35:37 I understand your reasoning and agree totally. 01:35:52 I wonder how difficult it would be to split out HAL, RTAPI, and emc 01:36:03 difficult, I imagine 01:36:15 there would have to be a very good reason to do that work 01:36:25 well - RTAPI is pretty stable 01:36:37 HAL is fairly stable (only needs updating when a component changes) 01:36:58 Having more than one distro out there makes any exclusive claims mute. 01:37:08 though HAL has changed more than emc proper lately 01:37:14 what do you mean by exclusive claims? 01:37:36 That "this distro" is the real EMC. 01:38:01 that's one reason for having "no distro", and emc as an application 01:38:12 SWPadnos: exactly right 01:38:14 like any of the Windows based controllers 01:38:17 it *is* an application 01:38:33 it's just very picky how it runs 01:38:35 It will never even approach being just another app. 01:38:51 rayh: wait until you install it on ubuntu before you think that 01:38:59 it's close now, actually - it just has more specific kernel requirements than most apps 01:39:00 rayh: I worked really hard to make it as easy as any other app. 01:39:11 You have not heard the horror stories I have already. 01:39:32 rayh: then how am I supposed to help? 01:39:48 You are doing great. 01:39:50 rayh: I've had some good bug reports (from jmk) and fixed the problems he ran into 01:40:15 I saw that. 01:40:25 rayh: I was hoping he would pop in tonight, to see how my latest fixes worked 01:40:33 there should be an ubuntu category for the SF tracker, maybe 01:40:57 I'd put it on the wiki. 01:41:05 Rather than making it a part of sf. 01:41:17 that's good for install info, but bug reports are a different story 01:41:36 here's where it gets confusing 01:41:48 I don't see how a ubuntu bug with finding a cdrom is an EMC specific problem. 01:42:02 If someone can't install ubuntu and get it to run BEFORE anything with emc2, there are other bug trackers more appropriate than theirs. 01:42:04 well - that's different ;) 01:42:05 rayh: exactly 01:42:13 err than OURS 01:42:30 And the BDI shouldered all of that burden. 01:42:42 right, and I believe that's the wrong way to go about it. 01:42:53 To make a single system that worked across a very wide range of machines. 01:43:00 install problems - those are generally rare for ubuntu - do you know the hardware these people had problems with? 01:43:16 You realize that you are talking to the inventor of the BDI? 01:43:21 rayh: jmk had to boot the kernel with "noapic" for his network card to work 01:43:53 I think the idea is that the emc developers should be concerned with emc, not necessarily the OS that it runs on 01:43:56 rayh: no, I didn't, but that doesn't change my opinion 01:44:20 argh, the message about noapic was for SWPadnos 01:44:24 though it's definitely helpful to provide a BDI-type install, and deal with tech support issues that arise from it 01:44:26 Sure. It is not the responsibility of the EMC core to make a full distribution. 01:44:33 I'm on my terrible laptop keyboard 01:44:45 Yuck. 01:45:05 regarding ubuntu - I know it's great with modern hardware, I'm not so sure about older hardware - the kind of machine that people would relegate to running emc (or that has an SIA slot) 01:45:16 ISA - duh 01:45:39 SWPadnos: there seems to be problems in recent 2.6 kernels that require fiddling with apic boot flags 01:45:42 my dyslexia didn't even notice. 01:45:50 heh 01:46:01 SWPadnos: I think that's fixed in my kernel, but in the installer kernel it's a problem. 01:46:08 sometimes - possibly for buggy BIOSes 01:46:12 There are huge problems with the entire 2.6 kernel and rtai or any other real-time system. 01:46:22 exactly (linux complains bitterly about those BIOSes) 01:46:36 linuS 01:46:38 I had no trouble on my dual Opteron machine (installing from a SATA DVD-RW to an SATA HD) 01:47:01 I have also not had problems, but jepler had trouble on his laptop, jmk on his desktop 01:47:03 I haven't been able to install Windows 2000 on that machine (even SP4), nor Gentoo 01:47:06 Try getting a 2.6 from HardHat or WindRiver or ... 01:47:07 both were fixed by twiddling apic boot flags 01:47:21 cool 01:47:32 I bet the next ubuntu release will be better 01:47:45 I saw on their wiki they are collecting data about the apic problem and intend to come up with a fix 01:47:58 rayh, the people you've talked to had problems with the initial install (like not getting through it, or not being able to reboot into ubuntu)? 01:48:01 but even now it's easy to work around 01:48:04 yep 01:48:30 Hardware issues mostly. 01:48:45 solvable, or did they just move on to BDI? 01:49:04 (and if so, did that work?) 01:49:13 Went back to BDI. 01:49:17 Most of em. 01:49:31 how many people have failed and how many succeeded? 01:49:32 That is why I was asking about the ease of making a BDI deb. 01:49:41 I don't expect paul to do it any time soon. 01:49:52 BDI4.40 is expected this summer 01:50:00 but I could set up a 4.38 and do the compile here. 01:50:06 he has also removed AXIS from his bdi because jepler wanted him to provide source 01:50:07 according to a message on the new mailing list 01:50:57 Would we get conflicts in the debian directory at sf if we try to make two versions? 01:50:59 it's probably not too difficult to make any needed changes to the .deb build info (whatever that file is called) 01:51:15 one would be under sarge/, the other under breezy/, so probably not 01:51:20 rayh: sort of - you would have a few files different to build it. 01:51:58 rayh: it would take a couple iterations to get right, but it probably wouldn't be too hard. 01:52:23 rayh: do you know when people get the brown screen? 01:52:29 I understand. Not for right now. I'll think more about it after the release of the source. 01:52:56 No I don't understand enough about the system to get past their frustration. 01:53:10 I do have a fellow overnighting me a disk. 01:53:32 well write down this for next time: at the first prompt instead of just hitting enter, type 'linux noapic' 01:53:47 I did. 01:53:52 I think that would have been the only thing jmk would have had to do and it would have been perfect 01:54:00 it's really similar to almost any other distro - except that it uses gnome by default, instead of KDE 01:54:01 We should build the kernel without. 01:54:06 I did 01:54:13 it usually works, and it's needed on some machines 01:54:13 but that doesn't help the install of course 01:54:28 kubuntu? 01:54:35 but probably not the ones that people would use for emc 01:54:42 kubuntu is KDE-based by default 01:55:10 didn't look like it was being maintained as well as ubuntu. 01:55:22 I think it's much newer 01:55:27 haven't looked at it myself 01:55:38 (all desktop environments look the same to me anyway) 01:55:43 In the long run, I've got my money on RedHat. 01:55:47 I installed the kubuntu packages after installing "normal" ubuntu 01:55:52 it works fine 01:56:06 $5 a share to $26 right now 01:56:15 rayh: it's true someone could build rpms just like my recent debs 01:56:15 But my heart is with SuSE. 01:56:35 I bought at $60 about 5 years ago (and almost made money) 01:57:06 Google at 60! 01:57:32 that's why the tech bubble burst you know - it was my $1000 investment 01:57:53 same with the snowless winter - I bought a snowblower 01:58:04 They saw you coming and timed it just after. 01:58:13 if you need rain, let me know - I'll leave my raincoat at home 01:58:16 Buy those in the spring. 01:58:22 SWPadnos: thanks for the winter 01:58:44 any time - but look out - when the snowblower breaks, it'll snow 01:59:16 SWPadnos: in about May, buy a new air conditioner please 01:59:23 That's the truth. Or when the neighbor's jeep is t-s up. 01:59:31 ok - that sounds like a good idea 01:59:39 our has been getting loud and not too cold 01:59:50 aha 02:00:47 Gotta get out of here, guys. Thanks. 02:00:51 hmmm - BDI doesn't have udev, right? 02:00:54 see you later 02:01:00 don't know 02:01:02 by ray 02:01:06 bye ray 02:01:14 nope no udev. it conflicts with rtai. 02:01:26 rayh: (I fixed that) 02:01:26 rayh has quit 02:01:27 not on ubuntu, I gather 02:01:37 no, all you have to do is do it right 02:01:53 I'm just looking at all the files in emc2/debian, to see what might have to change for BDI 02:02:16 some I haven't checked in, because they'll be wrong for other systems 02:02:37 udev and gnome menu entries come to mind 02:03:05 yep, plus the RT dir is /usr/realtime-2.6.12.6-magma 02:03:22 yes those dirs are different 02:04:24 ok - and some of the dependency package names 02:04:32 (like RTAI) 02:04:36 right 02:04:49 and not much else, it seems 02:05:04 probably not much else. Like I said it'll take a few tries. 02:05:31 well - I may take a stab at it, or I can test any changes you suggest 02:05:45 though I'm on BDI4.30 right now, not 4.38 02:06:08 I think ray still suggests people use 4.30 02:06:15 I don't suggest 02:06:25 so I have no idea what deb we'd build if we decide to do one. 02:06:30 yep, but I think it's due to the differences in installed packages 02:06:37 * cradek shrugs 02:06:45 not necessarily issues with 4.38 02:07:44 in theory, if I replace the sources file with the one line for these packages, and then install all necessary packages (emc, rati, kernel ...), it should work anyway 02:07:50 as you said - use them all as a set 02:08:53 hmmm - you know, we could just transfer the linuxcnc.org domain to my dreamhost account 02:09:15 I don't think it'll go over the transfer limit, and there would be no cost (until I need the bandwidth, which will probably be a while) 02:09:17 that's an interesting thought 02:10:21 I can add unlimited user accounts, email addresses, etc 02:10:32 there's also CVS/SVN support, and phpbb forums 02:10:34 I'm tempted to take you up on the offer to host my packages 02:10:35 etc etc 02:10:59 jmkasunich has joined #emc-devel 02:11:05 hi jmk! 02:11:06 do you have transfer stats for them? 02:11:09 hi jmk 02:11:10 you missed ray by 10 minutes 02:11:15 hi 02:11:25 SWPadnos: I don't understand the question 02:11:59 jmkasunich: google says you can put "linux noapic" at the initial ubuntu install boot prompt to fix the problems you had 02:12:00 is that "ray wanted to talk to me" (or "ray was talking about things that would interest me", or just "ray was here, now he's not"? 02:12:11 the middle one probably 02:12:19 we talked about packaging some 02:12:29 logger_devel: bookmark 02:12:29 See http://solaris.cs.utt.ro/irc/irc.freenode.net:6667/emcdevel/2006-02-10#T02-12-29 02:12:33 I go for door number three :) 02:13:04 cradek, I was wondering how much bandwidth has been used by people downloading the packages 02:13:06 axis even 02:13:14 (like the CVS tarballs and stuff) 02:13:28 SWPadnos: we would have to ask jepler about that 02:13:34 SWPadnos: I'm only taking about the debs though 02:14:08 the ones that are at dsplabs...ro 02:14:15 ok - I can ask alex 02:14:35 very few people are using the so far - the data would not be useful 02:14:41 true 02:15:25 theM 02:15:40 yO bOyZ! 02:15:52 this laptop kb is awful 02:16:06 I can't stand them - I actually carry a full size ketyboard when I travel 02:16:15 and I have a big laptop with a relatively nice keyboard 02:16:40 I only like my IBM Model M - not appropriate for a coffee shop 02:16:56 heh - much smaller than my 8500DV, I think 02:17:21 I mean my full-size keyboard is the IBM 02:17:35 but way too loud to use in a public place 02:17:37 ah - the old clicky keyboard with the 15-foot cord ;) 02:17:42 yeah that one 02:17:58 it has an interesting feature that my wife needed for her grad school 02:18:07 I have a stash of them, but I don't need it because they never break 02:18:13 it properly registers 6 keys being pressed at once 02:18:16 what's that? 02:18:18 haha 02:18:22 I'm not surprised 02:18:24 very good for brailling 02:18:59 most modern keyboards don't work, including my personal favorite, the Keytronic EO3600QLPS2 02:19:01 I have never understood why braille isn't typed on a normal keyboard 02:19:16 I guess some braille chars mean more than one letter (th) 02:19:20 (I think) 02:19:24 yep 02:19:31 position dependent as well 02:19:41 have not worked with braille for a long time 02:20:09 just on the drive-up ATMs ;) 02:21:08 funny 02:21:18 jmkasunich: so how are things? 02:22:13 * jmkasunich worked late 02:22:21 and now he's reading back in the log 02:22:30 you too, alex said he worked from 0900 to 2130 02:22:41 ray's not so hot on ubuntu is he? 02:23:06 I can't tell 02:23:19 he seems negative about it, maybe because of problems, but he won't say in detail what the problems are 02:23:25 he doesn't have it, and he's only heard problems 02:23:40 that makes it hard 02:23:43 yeah 02:23:54 why does he care at this point - its an experiment that you are trying 02:23:56 I sure don't know who is having the problems and calling ray 02:24:07 yeah - that is a puzzle 02:24:27 I'll help if I can, but I don't have the information to do it. 02:24:39 maybe in a few days it will be different 02:24:41 yeah, if you aren't willing to fumble thru (or contact the author, cradek), don't try it 02:24:51 I guess people think Ray is all things emc 02:24:53 well I NEED the bug reports 02:25:02 but only good ones :-) 02:25:17 hey - there's a bug 02:25:19 jmkasunich: we found several things the other night, and I think they are fixed now 02:25:22 (how's that) 02:25:34 which reminds me. 02:25:41 there are four available updates 02:25:43 SWPadnos: no, "there's a bug and I'm really frustrated" 02:25:46 emc, emc-axis, and two kernel 02:25:52 jmkasunich: right 02:25:55 I assume you'd like me to load them? 02:26:01 jmkasunich: yes please 02:26:09 jmkasunich: there should be a rtai update too 02:26:35 nope, 02:26:41 hmm 02:26:52 linux-headers-2.6.12-magma ver cjr9 02:27:04 everything should be cjr10 I think 02:27:09 try apt-get update 02:27:10 linux-image-2.6.12-magma ver cjr9 02:27:40 it only runs once a day I think 02:28:32 got them now 02:28:44 strange tho, apt-get update doesn't work 02:28:51 howso? 02:28:55 E: Could not open lock file /var/lib/apt/lists/lock - open (13 Permission denied) 02:28:55 E: Unable to lock the list directory 02:28:55 j 02:28:58 sudo 02:29:01 duh 02:29:14 better 02:29:17 kernel, rtai should be cjr10 02:29:24 emc2 and emc2-dev alpha17 02:29:35 emc2-axis 1.2a1-0.3 02:29:56 got cjr10, and emc2 alpha17 02:29:58 no -dev 02:30:01 jmkasunich: you should be able to remove your noapic kernel flag 02:30:09 maybe you have not installed emc2-dev 02:30:16 could be 02:30:32 is there a menu item for snaptic on ubuntu? 02:30:37 yes 02:30:42 where? 02:30:43 system/something/synapti 02:30:45 c 02:30:55 (not at my machine) 02:31:18 under system / package management 02:31:21 or something like that 02:31:26 not the applications menu 02:31:29 system/admin/synaptic maybe 02:31:43 yeah that's it 02:31:47 or maybe I added that to the menu :) 02:32:03 I just dragged it to the top panel 02:32:04 swp: I think you must have added it 02:32:09 nothing like that here 02:32:20 or Applications > Add Applications, then choose "Advanced" from the menu 02:32:25 hmmm - it has a pretty icon and everything 02:32:26 there is a system tools menu, but under it, no admin, and no synaptic 02:32:35 no admin?? 02:32:41 Package Management, not Synaptic 02:32:52 the second thing on the menu? Administration? 02:33:22 there is a "System Tools" menu, no administration on it 02:33:27 2nd item is bug reporter 02:34:03 system tools is next to the menu that has "logout" on it, right? 02:34:33 EBKAC 02:34:51 hahahahahaha 02:35:00 heh 02:35:02 there are three main menus, I was looking under System Tools in the first menu 02:35:39 god some users 02:35:50 thats why I use the cmd line 02:35:52 heh - and I see that all of the items I named were wrong :) 02:35:54 no ambiguity 02:36:20 except for stuff like "is that -p or -P" 02:36:27 or -pP 02:36:43 -p or -P works fine in email or IRC 02:37:29 I do wonder why ubuntu still uses firefox 1.07 instead of 1.5 02:37:31 ok, the kernel image install is asking me about a conflict with a pre-existing dir of the same name 02:37:48 jmkasunich: tell it to go ahead (I think the right answer is n) 02:38:00 right, it asks if I want to stop 02:38:05 right, n 02:38:14 (which is stupid) 02:38:55 not if they really really want you to stop, and expect people to reflexivly hit Y without reading ;-) 02:39:32 some wild messages on the kernel image install 02:39:45 the YOU MUST REBOOT thing? 02:39:49 yeah 02:39:54 yeah that's normal too 02:39:57 THEN REBOOT ALREADY! 02:40:01 I hate linux GUIs 02:40:05 I can avoid those by changing the kernel versions 02:40:12 but then the grub menu will fill up with crap 02:40:25 I tried to copy one of the messages from the terminal window 02:40:29 I hope to hash out the kernel issues before many users have it 02:40:29 but instead aborted the install 02:40:40 ctrl-c? 02:40:52 fscking ctrl-C should mean copy, or sig-intr, but not both at random!@ 02:41:06 I disagree, it should mean INTR always 02:41:07 all you have to do is amrk, then middle-click 02:41:12 mark 02:41:36 the installs are idempotent so you just have to run it again 02:41:39 ctrl-c/ctrl-v are commonly known (more so than middle click) and work often enough that when the don't it pisses people off 02:41:59 yeah they mean intr and page-down 02:41:59 idempotent? thats a $0.50 word for sure 02:42:03 :-) 02:42:04 I agree that it's weird, but expected in a terminal 02:42:32 (it's the fact that the terminal is in a GUI that makes it cofusing :) ) 02:42:38 argh 02:42:54 terminals use emacs keys, apps nowadays use windows keys, to make it equally confusing for everybody 02:43:17 if only we could go back to the wrodstar (and turbo C) days 02:43:20 gah 02:43:22 wordstar 02:43:31 ctrl-K ctrl-K 02:43:51 jmkasunich: I learned that word reading the debian policy guide 02:44:18 I just got used to using the edit/paste menu in ksirc, because ctrl-v only works sometimes there 02:44:27 now in xchat there is no edit/paste 02:44:37 so its back to ctrl-v 02:45:08 I just hate change, thats all 02:45:24 no emc2 for you - you have to keep emc(1) 02:45:32 lol 02:45:47 but first, you have to write a punched-tape driver for it 02:45:59 fsck that 02:46:08 that's easy, my punch just plugs into the serial port 02:46:35 does it use RTS/CTS to pause the tape? 02:46:51 how much pausing do you need at 110 baud? 02:46:55 heh 02:46:58 hmm, the terminal window says "setting up emc2-axis (1.2a1-0.3) ... 02:47:00 then nothing 02:47:07 how many bytes do you need for a 1 minute move 02:47:11 a message outside the terminal says its all done tho 02:47:13 is it done? look above 02:47:20 yeah it's done then 02:47:46 would be less confusing if the emc-axis install printed _something_ 02:48:03 packages typically don't print much 02:48:07 normally you'd get a command prompt and know it's done 02:48:08 I know some packages print a dictionary, some don't print much 02:48:10 if it's in a real terminal, you get a prompt 02:48:31 well anyway, its done 02:48:31 it would be nice if they had an "all done" message in their debug terminal 02:48:36 yay 02:48:41 so time to reboot 02:48:43 be sure to reboot before you run emc 02:48:44 bye! 02:48:46 bye 02:48:49 hey 02:48:50 yep, though you can check the "close on successful completion" checkbox 02:48:55 not yet 02:48:57 don't forget you can remove your noapic flag 02:48:59 gotta remove the noapic 02:49:02 heh 02:49:08 it probably did it for you :-/ 02:49:48 yeah, and it put in the damned splash too 02:50:02 I hate bootsplash 02:50:09 I want to see the gears turning 02:50:18 not the chrome 02:50:50 you should get some chrome gears with the graphical startup 02:51:02 :-P 02:51:20 at leasst I get a half-screen of messages (starting netwrorking [ok]), and the kubuntu logo above 02:51:31 not hte full boot messages though 02:51:38 I need to lkearn to type 02:51:40 see 02:52:19 jmkasunich: you can change the automagic defaults 02:52:24 doing that now 02:52:29 nonaltoptions= 02:52:30 jmkasunich: there are instructions in menu.lst 02:52:59 I'm less than fond of that automagic stuff 02:53:06 comments that aren't, thats just wrong 02:53:07 jmkasunich: you get to see the new axis feature, colored backplot 02:53:19 yeah having magic comments is total crap 02:53:26 oooh - G0/G1/G2/G3 are different colors now? 02:53:29 yes 02:53:32 cool! 02:53:48 is that in the latest tarball? 02:53:51 yes 02:54:01 when was it added? 02:54:03 that info is passed out of the TP into the stat buffer now 02:54:03 rebooting 02:54:07 jmkasunich has quit 02:54:25 SWPadnos: I did the emc part a while back, but the AXIS part only a couple days ago 02:55:08 ok - I've got 20060204, so I probably need an update 02:55:52 looks like it went in on 0206 02:56:08 sometimes it's a fast-moving target 02:56:13 ok - I didn't rememner seeing it 02:56:26 r e m e m b e r 02:56:54 do you read our changelogs? 02:56:59 I don't think it got any fanfare 02:57:08 only when prompted on IRC 02:57:50 jmkasunich has joined #emc-devel 02:57:51 by the way - what kind of bandwidth / transfer stats do you have for the axis stuff? 02:58:02 well the kernel seems to work 02:58:18 great 02:58:25 without special flags? 02:58:34 SWPadnos: the axis website is just on my home dsl 02:58:40 no special flags 02:58:41 SWPadnos: it's only a few hundred megs a month, if that 02:58:44 ok 02:58:50 what's your upload speed? 02:58:54 384kbps 02:59:05 ok 02:59:15 green is rapid, red is cutting... nice 02:59:23 SWPadnos: http://unpy.net/usage_axis/usage_200601.html 02:59:51 850M - thanks 03:00:00 SWPadnos: I actually looked into moving to hosting, but the python CGI scripts I use to generate the pages are so slow on a shared machine that it was a step backwards 03:00:27 ah. I was thinking of just mirroring things on my dreamhost account 03:00:39 we can add / move domains there if desired as well 03:00:48 it was actually dreamhost that I was using 03:00:50 you know what really sold me on your kernel? the fact that time measurement works1 03:01:04 I don't need the bandwidth for a while, and I get 1.8TB/month 03:01:20 yeah, they sure have plenty of bandwidth and are pretty generous with the disk space too 03:01:28 yep 03:02:00 and it keeps going up, at no charge 03:02:47 I like the shell access as well (plus CVS, SVN, and other good stuff) 03:02:52 jmkasunich: does it still work? 03:03:24 jmkasunich: btw arcs are a different color too 03:07:08 hmmm. it appears that I can't have the same version of axis working with more than one emc2 install (rip, actually) 03:07:32 SWPadnos: you get an error? 03:07:48 SWPadnos: if they're different versions of axis, it might lead to trouble 03:07:56 well - I did, and I ignored it because I was about to download a neew version anyway 03:08:00 one sec 03:08:26 emc2-axis deb plus axis for rip emc2 are probably trouble 03:08:44 ok - just the standard "can't execute DISPLAY program - I may have done a make clean and not rebuilt axis again 03:08:59 this is rip only, no installed emc2 03:09:05 yeah make clean removes bin/* which is terrible 03:09:16 ok - that's the problem - nevermind 03:09:20 oh yeah, there is that emc2 .. feature 03:09:26 I was testing the halconfig changes and I'm sure I did make clean 03:12:16 hmmm - should the backplot get cleared automatically when you load a new file? 03:12:32 cradek: yes, still working (was reading mail) 03:12:53 I guess 3dchips has no arcs, I only saw green (rapids) and red (cutting) 03:13:01 yeah cds has some 03:13:14 but you can take my word for it I bet 03:13:31 yep 03:13:32 I just noticed that, and loaded cds, which prompted my backplot question :) 03:13:47 04 s32 RW 9671 ddt.0.tmax 03:13:47 04 s32 RW 9883 ddt.1.tmax 03:13:47 04 s32 RW 9005 ddt.2.tmax 03:13:47 04 s32 RW 9053 ddt.3.tmax 03:13:47 04 s32 RW 12068 ddt.4.tmax 03:13:48 04 s32 RW 9199 ddt.5.tmax 03:13:48 what question? 03:13:50 02 s32 RW 27300 motion-command-handler.tmax 03:13:52 02 s32 RW 56521 motion-controller.tmax 03:14:00 hmmm - should the backplot get cleared automatically when you load a new file? 03:14:02 56 us to run the motion controller code, not bad 03:14:45 can you try stepgen? 03:14:47 (thats the max during the entire run of chips, average seems more like 12-14uS 03:15:08 I'll try the stepper now 03:15:12 so you do get good timing info still 03:15:14 that's great 03:15:32 that alone will move me personally off of BDI 03:15:45 the timing is busted on pauls kernels 03:15:52 I'm curious as to why there's been a recent rash of people not being able to get reasonable BASE_PERIOD settings to work 03:16:13 SWPadnos: since there is no testsuite on bdi, it's impossible to tell what's up 03:16:16 (if that's been their problems) 03:16:20 true enough 03:16:32 and since the timing doesn't work, you can't check that 03:16:47 right - that's why I asked ;) 03:17:12 damn, joint 2 following error 03:17:32 jmkasunich: so for my kernel, can you do the missed deadline checks? 03:17:33 ok - that's what Gene was getting 03:17:46 yes, I think 03:18:09 jmkasunich: on what program? 03:18:16 3dchips 03:18:30 is it the initial g0->g1 move at the beginning? 03:18:33 just as a foot was starting? 03:18:46 didn't have backplot on 03:19:00 the code window has line N100 highlighted 03:19:14 but is that the readahead spot? 03:19:32 no, that's the running line +-1 03:19:59 incidentally, I have to comment out line N50 to get axis to run the file 03:20:09 what's N50? 03:20:10 tkemc works though 03:20:18 sim-axis ran it fine here 03:20:19 N50T1M6 03:20:39 did you get an error? 03:20:40 I'll check that again, just to be sure 03:20:56 no, it just doesn't run past that line - not crashed, but hung 03:21:10 where are the installed ini files? I want to check the accel limits in stepper.ini 03:21:15 hung exactly how? 03:21:20 jmkasunich: /etc/emc2 03:22:04 one sec 03:23:03 ok - no hilighted line, no tool motion, manual controls are grayed as normal 03:23:34 pause / play / stop buttons work as expected, and I can move the backplot around 03:23:43 hmm, second attempt to run 3dchips had no ferror 03:23:45 anything on stdout? 03:23:50 but play and continue have no effect - it just doesn't run 03:24:21 jmkasunich: that's a problem alex has also seen 03:24:32 the comments are output, then nothing until I press stop, when I get "EMC_TOOL_ABORT" 03:24:35 I have a theory 03:24:36 jmkasunich: bad blend probably 03:25:13 direction reversal at rapid 03:25:14 jmkasunich: what's your theory? 03:25:36 but at N100->N110 there is no reversal 03:25:41 colinear g0->g1 03:25:44 line 80 is the first move, no Z component 03:25:49 line 90 goes to Z10 03:25:57 line 100 goes to Z-56 03:26:01 that's true 03:26:08 it starts at Z0 03:26:15 so 90 goes up, and 100 reverses 03:26:16 what was your z coord when it ferrored? 03:26:22 9.2 03:26:23 that would tell us which blend 03:26:27 just beginning the down 03:26:32 ah 03:26:33 or about to 03:26:41 yeah it could have been going up still 03:26:45 if I reload and restart, it is OK, because the +10 move doesn't happen 03:26:51 gonna scope it now 03:27:15 this works with univstep, so I'd have to suspect stepgen 03:27:34 probably the config (not enough headroom or something) 03:27:51 but could be stepgen's "pretuned" position loop isn't 03:28:02 I think the default is 20 / 21 for accel, and 1.2/1.2 5 for vel 03:28:10 1.2/1.25 03:28:15 20/21, and 1.2/1.4 03:28:16 I had problems similar to this with my etch-a-sketch. They started happening consistently when I turned on backlash compensation 03:28:35 right 03:28:38 in halscope the accels and velocities during backlash compensation (after a reversal) were crazy, way out of bounds 03:28:51 I don't have any screenshots, unfortunately. 03:29:07 here's an ascii art screenshot: 03:29:14 ___/| 03:29:30 something like that? 03:30:13 If I recall correctly, the accel spiked to +large then -large, and there were "impulses" on the acceleration graphs: ______|______ 03:30:24 ah 03:31:09 I think this only happened when I entered nonzero numbers for backlash 03:31:17 backlash has issues 03:31:23 no accel is applied, for one 03:32:32 the ferror is occurring as Z reverses from rapid up to rapid down 03:32:34 I think backlash could be done before canon - add a "segment" and let TP plan it 03:32:46 eww 03:32:52 haha 03:32:57 let's not worry about that now 03:33:02 good 03:33:03 just preprocess the whole file 03:33:17 later I can explain why I don't think backlash should be in canon 03:33:30 backlash shouldn't be in the machine at all, but oh well :) 03:33:46 let's pretend I didn't say anything about that 03:33:55 the backlash is killing you 03:34:05 ack 03:34:21 jmkasunich: is the rapid reverse problem in tp or stepgen? 03:34:26 SWP: I think it was you that fixed the offset entry box in halscope 03:34:35 heh - my wife's cell phone decided to flip the outside display vertically - I almost thought it was back to normal because the time was 10:33 03:34:36 thank you, thank you, its so much nicer now 03:34:48 no - you did it while we were on irc 03:34:51 cradek, looking now 03:34:52 so thank you instead 03:35:04 well you splained it to me, or something 03:35:09 something like that 03:35:29 I think you found the fix while I was looking for the problem ;) 03:35:42 darn, wish I had the ddt blocks in this config 03:35:50 maybe I'll add them 03:35:56 yeah those are nice 03:36:04 reading velocity off the scope screen isn't hard, but accel gets tricky 03:36:16 actually.. this isnt that tricky 03:36:58 prior to the fault, vel is ~0.022" in 20mS 03:37:16 maybe that 0.024, which would be 1.2ips 03:37:19 is that commanded vel? 03:37:22 yes 03:37:35 (and fb as well, its tracking well at that point) 03:37:51 then vel goes to zero in about 20mS 03:37:56 I hope it's not the tp I hope it's not the tp 03:38:20 accel limit is 20 ips/s 03:38:32 so 1.2ips should take about 60mS 03:38:43 ouch 03:38:49 if I'm doing the math right 03:39:07 do comand/fb split wildly there? 03:39:11 m 03:39:12 yeah 03:39:28 you guys know any convenient place to post a screenshot? 03:39:40 do you have debug output that you can check for the last SET_VEL message? 03:39:45 err SET_ACCEL 03:40:00 I ran it from the icon, so no debug output 03:40:06 darn 03:40:12 but I could run it again 03:40:16 it might be in your ~/.xsession-errors 03:40:26 pastebin is only for text, right? 03:40:31 (want to post the scope pic) 03:40:41 dcc it to me, I'll post it 03:41:20 hmm, that nice "take screenshot" command grabs the whole screen, not a window 03:41:26 * jmkasunich maximizes scope ;-) 03:41:33 haha 03:41:40 there's a key for taking a snap of a window too 03:41:52 you might have to set it in system/preferences/key shortcuts 03:42:30 cool, thanks 03:42:45 (you didn't really want a 1600x1200, did you) 03:42:51 nope 03:43:02 now I just have to figgure out this dcc stuff 03:43:04 I have 3200x1200 at home, 640x480 here 03:43:39 "/dcc-send cradek" 03:43:45 I think ou can rightclick on me, "send file" 03:43:45 y 03:43:48 you should get a file dialog 03:44:14 hmmm - chatzilla doesn't do that. I should file a feature request 03:44:34 did you get the offer? 03:44:46 yes, but something on your end is blocking the transfer 03:44:50 nat/firewall? 03:45:01 could be, I have a NAT router 03:45:11 yeah you have to jump through hoops for dcc with nat 03:45:21 strange - I didn't 03:45:28 ok, I think the hoops for putting it on my own web space may be lower 03:45:31 your system tells mine "contact me on port xyz to get this file" 03:45:41 just have to remember an obscure URL 03:45:49 but port xyz doesn't get back to your machine 03:48:11 http://home.att.net/~jmkasunich/Pics/Screenshot-HAL_Oscilliscope.png 03:48:30 green is commanded pos, red is feedback, both at 0.020"/div 03:49:05 other red is following error at 0.005" / div 03:49:58 I need to add cursors to the scope 03:50:09 ah I see it now 03:50:48 green stops and actually starts going back down a tiny bit, while red is still decelling 03:51:07 one day I'll fix that spelling error ;) 03:51:09 yeah the red is way behind 03:51:13 if I did the math right, red is correct and green deceled at 60 ips/s 03:51:44 did you check your .xsession-errors 03:51:56 I would like to verify the SET_ACCEL 03:52:35 damn some people don't know how loud they are when talking about private things in public 03:52:43 ? 03:52:49 I'm at a coffee shop 03:53:02 and I'm learning a whole history of this couple behind me 03:53:14 crapload of gnome-cups-icon warnings and stuff, emc messages are lost in the noise 03:53:33 darn 03:53:49 grep says no ACCEL in there 03:54:02 no SET for that matter 03:54:09 bet debug is set low 03:54:11 you must not have that much debug on 03:54:26 ok, I'm gonna change debug and start from a terminal 03:55:18 yeah, debug was 1 03:55:29 should I go full blast, 7FFFFFFFF 03:55:38 that's probably good for the debs, since you don't want to fill up logs 03:55:41 yeah go all aout 03:55:47 out 03:56:09 some time I should figure out what debug gives just the canon commands 03:56:12 that's usually what I want 03:56:43 damn, I should have done tee or something 03:56:49 wanna be able to grep the output 03:59:47 john@ke-main-ubuntu:~$ cat log1 | grep SET 03:59:47 Issuing EMC_TRAJ_SET_SCALE -- (+209,+20, +1,1.000000,) 03:59:47 Issuing EMC_TASK_SET_MODE -- (+504,+16, +2, +2,) 03:59:47 Issuing EMC_TASK_SET_STATE -- (+505,+16, +5, +2,) 03:59:47 Issuing EMC_TASK_SET_STATE -- (+505,+16, +6, +4,) 03:59:48 Issuing EMC_TRAJ_SET_VELOCITY -- (+205,+28, +0,1.650446,1.650446,) 03:59:50 Issuing EMC_TRAJ_SET_ACCELERATION -- (+206,+20, +0,20.000000,) 03:59:51 are the Z and TRAJ accels the same in this config? 03:59:52 Issuing EMC_TRAJ_SET_VELOCITY -- (+205,+28, +0,1.200000,1.200000,) 03:59:54 I 03:59:56 those are the first few 04:00:00 (only set accel) 04:00:01 ok there's the 20 04:00:08 lemme check 04:00:50 traj: default 15, max 20 04:00:53 axis: max 20 04:00:57 ok 04:01:17 wait is there no SET_ACCEL before the first move? 04:01:17 velocity: traj and axis max vel both 1.2 04:01:37 wouldn't that be something 04:01:55 the first move is x and y tho 04:02:02 the grep probably hits things 04:02:10 lemme grep for "Issuing" 04:03:04 john@ke-main-ubuntu:~$ cat log1 | grep Issuing 04:03:04 Issuing EMC_TRAJ_SET_SCALE -- (+209,+20, +1,1.000000,) 04:03:04 Issuing EMC_TASK_SET_MODE -- (+504,+16, +2, +2,) 04:03:04 Issuing EMC_TASK_ABORT -- (+503,+12, +3,) 04:03:04 Issuing EMC_TASK_PLAN_OPEN -- (+506,+268, +4,/usr/share/emc/ncfiles/3D_Chips.ngc,) 04:03:06 Issuing EMC_TASK_SET_STATE -- (+505,+16, +5, +2,) 04:03:08 Issuing EMC_TASK_PLAN_SYNCH -- (+516,+12, +0,) 04:03:10 Issuing EMC_TASK_SET_STATE -- (+505,+16, +6, +4,) 04:03:12 Issuing EMC_TASK_PLAN_RUN -- (+507,+16, +7, +0,) 04:03:14 Issuing EMC_TOOL_PREPARE -- (+1104,+16, +0, +1,) 04:03:16 Issuing EMC_TOOL_LOAD -- (+1105,+12, +0,) 04:03:18 Issuing EMC_COOLANT_FLOOD_ON -- (+1406,+12, +0,) 04:03:20 Issuing EMC_SPINDLE_ON -- (+1304,+20, +0,1600.000000,) 04:03:22 Issuing EMC_TRAJ_SET_VELOCITY -- (+205,+28, +0,1.650446,1.650446,) 04:03:24 Issuing EMC_TRAJ_SET_ACCELERATION -- (+206,+20, +0,20.000000,) 04:03:26 Issuing EMC_TRAJ_LINEAR_MOVE -- (+220,+64, +0,2.086614,-2.209764,0.000000,0.000000,0.000000,0.000000, +1,) 04:03:29 Issuing EMC_TRAJ_SET_VELOCITY -- (+205,+28, +0,1.200000,1.200000,) 04:03:31 I 04:03:36 the first linear move is issued after the set vel and set accel 04:03:41 ok 04:03:49 that means it's not something easy 04:03:55 darn 04:04:10 crap, 11 pm and I still haven't started on the compile farm problem 04:04:11 the first move must be diagonal in xy 04:04:14 yes 04:04:22 then up Z, then down Z 04:04:25 sigh 04:04:48 I could block blending for a reversal, but that won't fix the almost-reversal problem 04:05:04 I wish someone would come out of the woodwork and take simple_tp and run with it 04:05:11 question - why is the first set-velocity setting it to 1.6? 04:05:14 the limit is 1.2 04:05:18 that's 1.2 sqrt 2 04:05:20 I bet 04:05:25 because it's diagonal 04:05:36 ok 04:05:58 but later the move is in one axis so it has to be slowed down to 1.2 04:06:10 here are the Z moves 04:06:11 Issuing EMC_TRAJ_SET_VELOCITY -- (+205,+28, +0,1.200000,1.200000,) 04:06:11 Issuing EMC_TRAJ_LINEAR_MOVE -- (+220,+64, +0,2.086614,-2.209764,0.393701,0.000000,0.000000,0.000000, +1,) 04:06:11 Issuing EMC_TRAJ_LINEAR_MOVE -- (+220,+64, +0,2.086614,-2.209764,-0.998898,0.000000,0.000000,0.000000, +1,) 04:06:11 I 04:06:31 up to +10mm (.393... inches), then down to -50mm 04:06:40 sure looks right 04:06:46 darn 04:07:02 so it's probably tp 04:07:14 I'm just glad its not stepgen ;-/ 04:07:22 I sure wish it was stepgen 04:07:24 or canon 04:07:27 although that would probalby be easier to fix 04:07:31 right 04:07:51 well I want to deal with the compile farm 04:07:57 ok, and I want to go home 04:08:03 should we/I submit a bug report for this? 04:08:03 talk to you over the weekend maybe 04:08:18 I have a bit more info on the axis problem, if you want it 04:08:22 it seems to be repeatable 04:08:24 hardly any use, we already know blends are wrong 04:08:35 but if you want to, it's fine with me 04:08:52 SWPadnos: email it? I'm done for the night 04:08:56 well it kinda sucks that one of our stock configs fails with out favorite demo g-code file 04:09:01 ok - I'll try to make it sensible 04:09:05 jmkasunich: agreed 04:09:15 fix the file? change the config? 04:09:30 jmkasunich: we could put a G4P.001 and it'll fix it 04:09:39 (IOW, sweep it under a rug until we have a fix) 04:09:43 right 04:09:45 why would it work fine using univstep though? 04:09:56 SWPadnos: more slop in the pid 04:10:03 stepgen rigidly enforces the accel limits 04:10:05 SWPadnos: if you plot your ferror, I bet you'll see it 04:10:13 it won't allow the TP to cheat 04:10:16 ok, maybe tomorrow ;) 04:10:28 but stepgen is getting higher limits 04:10:41 that would solve it 04:10:41 but not high enough for a misblend - got it 04:10:47 goodnight guys 04:10:48 s/solve/hide/ 04:10:51 night 04:10:55 night cradek 04:11:39 judging by the scope traces, the error would wind up at 0.050 or more 04:11:48 opening up the error limits won't fix that 04:12:20 hmm, it really is a TP error, I just realized that it never reaches the target position 04:12:34 the green trace? 04:12:49 its supposed to go to 0.393701 04:13:09 it reverses shortly after reaching 0.373 04:13:27 there are two red traces - what are they? 04:13:38 (scope offset is 0.393, so center of screen is 0.393 and one div down is 0.373 04:13:46 the one that tracks the command is feedback 04:13:50 (command is green) 04:13:53 ah - ok 04:13:59 the other one is error, at 0.005"/div 04:14:10 tripped at 0.010 error 04:14:17 yep 04:14:39 you know, IMO, blending should be off on rapid moces 04:14:41 moves evenb 04:14:45 even even 04:15:03 I think so as well, but Ray didn't (we had a short discussion on that topic) 04:15:16 rapid->rapid, regular->rapid, and rapid->regular should be exact stop 04:15:30 a friend of mine has issues with DeskCNC as well - rapid out of a hole, and it blends with the move to the next hole 04:15:45 I heard a similar complaint about emc 04:15:55 so he gets a gouge in the hole, unless he sets his clearance plane very high above the work 04:16:00 yep 04:16:32 though rapid->rapid should be blendable - rapids aren't even guaranteed to be coordinated 04:17:21 since blending seems to be busted, I'm really tempted to turn it off for any rapid move 04:17:27 turn it back on once fixed 04:17:44 hack hack, cough cough, wheeze ;) 04:17:46 in the meantime, like you say, rapids aren't guaranteed to be anything (including blended ;-) 04:18:13 what's G4 again? 04:18:22 (the G4P0.001) 04:18:27 dunno 04:18:31 heh 04:18:50 P0.001 means pause 1mS, right? 04:19:07 yep - dwell time 04:19:43 heh, I know one way to fix the 3dchips demo 04:19:47 swap two lines 04:19:57 put the Z up move before the lateral move 04:20:01 hmmm - it's only invalid if the P number is negative. I winder if G4P0 would just make everything exact stop mode? 04:20:08 so instead of lateral, up, down, it would be up, lateral, down 04:20:32 up lateral down would make more sense anyway, retract before move so you don't break a tool 04:20:53 the demo files aren't exactly runnable in that regard 04:21:25 cds starts with the tool below the work, for instance (I think) 04:21:45 well its a good excuse to swap the lines anyway 04:21:56 kate 04:22:04 wrong window 04:22:15 and no kate on this box anyway 04:22:17 gotta fix that 04:22:18 yeah - I'm Steve 04:22:26 that'll be a long download 04:22:43 you may want to try gedit ;) 04:22:49 don't want all of kde, just kate 04:23:01 gotta have kde-base and kde-libs, I think 04:23:14 not tonight then 04:23:41 easy enough to check, now that you know where synaptic is ;) 04:24:50 10 additional debs 04:24:55 wow 0 15 dependencies, 5 suggests, and 1 recommends 04:24:59 22Meg 04:25:10 not bad for an editor (har har) 04:25:35 at least that's the tally on BDI 4.3 04:25:45 gedit did the job 04:25:57 but I want a nice multi-file editor 04:25:59 it even hilights 04:26:01 later 04:26:10 it = gedit? 04:26:14 yep 04:26:18 kate does too 04:26:34 I know - just pointing out that gedit isn't crap, just not the same :) 04:26:45 yeah, I see that it has multiple files too 04:27:03 dunno if I can split the window tho 04:27:20 don't know about that 04:28:01 doesn't look that way, you can move a file to a new window, but not split the window to display two files (or two parts of the same file) 04:29:20 right 04:29:24 bummer 04:30:09 of course, I did add kate (and kde, for that matter) 04:30:26 the kubuntu packages are available in synaptic 04:30:34 though the desktop is still gnome 04:31:04 ok, the edits to the NC file fixed the ferror 04:31:35 IU'm not sure that's a good idea 04:31:49 why not? 04:31:59 you don't want to have people think that everything's great, only to have it fail on their own programs 04:32:17 it's not ideal to have them think it sucks, and move on to something else either though 04:32:30 well I can't fix it right now, and I certainly don't want 500 fscking folks asking the same question on the users list 04:32:49 "why do I get a following error on 3dchips" 04:33:06 those 500 people will be pissed off and still ask the question when their own programs give them following errors (and 3dchips doesn't) 04:33:19 yeah 04:34:02 except most of their programs won't rapid one direction then back the other direction on the very next move 04:34:06 leave the problem there, and some kind sould pissed off programmer might fix the TP ;) 04:34:15 soul, not sould 04:34:51 dammit 04:35:03 ? 04:35:19 you're probably right 04:35:23 grrrr 04:35:24 sorry :( 04:35:38 submitting a bug report 04:36:28 do you know what the rough tuning parameters in stepgen are (expressed as P,I,D ...)? 04:36:46 no, it doesn't use PID 04:37:15 the better way to fix the problem would be to increase the headroom (or decrease the max_vel/accel) 04:37:22 until the TP is fixed 04:37:32 the TP is asking for about 3x the proper accel 04:37:40 though I think Gene Heskett was fiddling with that 04:37:47 yep - that's about what he found 04:38:14 if we allow such extreme over-limit, either they have to really set the TP limits low, or they risk lost steps when the TP accels too fast 04:38:36 lost steps is worse than following errors 04:38:49 yep. damn 04:39:02 IMHO you should be able to set your stepgen limits at the motor's limits 04:39:18 as determined by sending a square wave into stepgen and letting it limit the accel and vel 04:39:42 correct, but then you may have to dcrease the limits for TP, just in case 04:39:46 decrease 04:39:56 then derate a little to set the TP limits and be good to go, knowing that no matter how stupid the TP and/or backlash comp code is, the stepgen won't let you over-accel or over-vel your motor and lose steps 04:40:37 I think that's the idea, the headroom is meant to be above the TP, but not above the motor capabilities 04:40:43 right 04:41:30 but when the TP sometimes goes 3x over its "limit", you either have to set the limit really low, or you will exceed the motor capability (or stepgen will enforce a limit and you will get ferror) 04:41:52 right, or you fix the TP 04:41:58 (but not tonight) 04:42:02 right 04:43:35 wait a sec - that cmd_pos trace was supposed to get to 0.373, or 0.393? 04:43:47 .393701 04:43:50 aka 10mm 04:43:54 right 04:44:04 only got to .373 and a bit 04:44:10 but in blend mode, it doesn't have to get there 04:44:17 it's trying to maintain speed 04:44:31 heh, maintain speed as you reverse direction? 04:44:38 hard to do -true 04:45:59 where is this code? src/emc/motion? 04:46:12 yes 04:46:16 I think 04:46:29 or src/emc/kinematics/{tp,tc} 04:46:33 kins 04:48:09 I'm not sure that the TP knows about G0 or G1 - only the feedrate is different 04:48:40 right, its just speed 04:48:49 it misblends everything, not just rapids 04:49:00 right 04:49:06 but rapids are more likely to be near the HW limit (stepgen or whatever) so you get ferror 04:49:10 thinking about the "no blending for G0" idea 04:49:24 TP knows about g0/g1 now 04:49:28 not the right solution, just a bandaid 04:49:29 motion_type 04:49:34 ok 04:50:20 it would be easy to not blend rapids 04:50:26 I do ned to figure out how to use screen ;) 04:50:29 need 04:50:42 I use screen constantly 04:50:46 I see that 04:51:22 oh, I see what you mean 04:51:26 heh 04:51:48 do you use the X version, or just the terminal? 04:51:58 terminal 04:52:06 I'm stuck in about 1994 04:52:36 hmmm - 1994. the land before "32-bit windows for the masses" :) 04:53:49 if you want to not blend for g0, you would put it in canon (before tp) 04:54:46 possibly. I wonder if the TP should consider a move done when it starts decel? 04:54:52 (as it does now) 04:55:13 not sure what you mean by done 04:55:22 I think during the blend both moves are marked active 04:55:52 there's a comment (though I didn't check the code) to the effect that a move is considered finished when decel starts, in blend mode 04:56:17 err - nevermind 04:56:24 damn, it automatically assigned the bug to me - thats not fair! 04:56:26 the next move begins, not the current move ends 04:56:29 heh 04:56:30 jmkasunich: haha 04:56:52 SWPadnos: right, the current move is marked decel, next move is accel, both are active 04:57:19 ok, so when decel is in the same direction as accel, there's a problem 04:57:26 the two values are added 04:57:34 I'm sure that's not the only problem 04:57:49 that would explain 2x accel limit, I'm pretty sure I saw 3x 04:58:05 I should probably put the ddt 04:58:12 blocks in to verify that 04:58:19 probably not the only problem - check lines 637 to 654 of tp.c 04:58:34 its too late for me to get into the code tonight 04:58:46 I've been trying to get more sleep, still recovering from a cold 04:58:53 bedtime very soon 04:58:56 heh - bummer 04:59:04 no 3AM sessions this week huh? 04:59:21 just remembered something tho, sim has ddts, and it should still show the problem (just not fault on it 05:02:37 ok, according to the ddts, the accel is exactly twice the limit 05:02:43 aha 05:03:18 oh - good 05:03:23 which aha? 05:03:31 sim is pretty nice for this 05:03:33 you're right about where it's wrong 05:03:42 just trigger on accel < -22 ips^2 05:03:59 woohoo! 05:04:36 I suspect the preAMax calculation that paul took out was related to this 05:04:36 I bet the blend time is calced as min(time from current vel to zero, time from zero to target vel) 05:04:44 which is only valid when both vels have the same sign 05:05:05 it's not actually calculated at all, it seems 05:05:12 oh, thats right 05:05:23 it actually calcs the independent moves 05:05:23 yeah also the blend time is very wrong if the accels are different 05:05:28 get accel, then scale accel (and add it in again) 05:05:38 this doesn't blend, it only uses the current move 05:05:39 (which include vel to zero and zero to vel ramps) 05:06:00 everything is thisTC, no nextTC or anything 05:06:40 anyway, it is trivially duplicated, which is half the battle IMO 05:06:45 maybe thisTc needs to be thisTc->next somewhere in there? 05:07:10 no you're inside a loop over the tc queue 05:07:31 I'll leave you guys to it, I gotta get my sleep 05:07:33 good luck 05:07:38 with if(done) continue; at the top 05:07:38 thanks 05:07:44 goodnight jmk 05:08:14 jmkasunich has quit 05:08:28 I don't see done, was that a paraphrase? 05:08:34 yes, sorry 05:08:44 line 542 05:08:49 oops - tcIsDone 05:09:16 I'll look at this tomorrow... it's too late tonight 05:09:22 I agree 05:09:29 let me know if you make any discoveries :-) 05:09:36 talk to you tomorrow 05:09:42 I think I'll discover how warm my electric blanket is 05:09:46 see you :) 06:04:44 SWPadnos is now known as SWP_Away 14:23:29 SWP_Away is now known as SWPadnos 18:03:57 steve_stallings has joined #emc-devel 22:15:47 logger_devel has joined #emc-devel 22:15:47 topic is: "Welcome to the Enhanced Machine Control development place. | Regular Developers' meetings 24/7 !" 22:15:47 Users on #emc-devel: logger_devel steve_stallings jepler cradek jtr_ anonimasu alex_joni LawrenceG @ChanServ SWPadnos 23:25:22 in case this reassures anyone of my good intentions, I'm installing bdi-live in qemu right now and will work this weekend to get emc2 compiling on it again