#emc | Logs for 2005-12-09

[00:00:41] <fenn> not sure that those limits would apply if you use a different cutter.. but they should be close
[00:01:21] <anonimasu> hm, they should
[00:01:21] <fenn> same problem as before
[00:01:22] <anonimasu> they are machine limits..
[00:01:32] <fenn> but the path is different
[00:01:40] <lerman> got the email. I don't see acceleration and velocity (at least in direct terms).
[00:02:11] <petev> no, they are in the formulas, but u can use the linear/circular work sheets
[00:02:26] <anonimasu> how different?
[00:02:34] <petev> the required accel or torque is what is calced
[00:02:44] <fenn> oh about 0.1mm or so i guess
[00:02:56] <anonimasu> gigantic :)
[00:03:14] <fenn> yeah, that's why i think its silly to worry about cutter comp affecting the path
[00:03:28] <anonimasu> actually that's lots..
[00:03:39] <anonimasu> I am not joking :/
[00:03:51] <petev> fenn: that's 4 mil
[00:03:59] <anonimasu> but still, how much difference does it make when machining
[00:04:02] <fenn> in terms of accel and velocity limits it's not much
[00:04:10] <anonimasu> ah yeah
[00:04:33] <anonimasu> it might be non optimcal not to recalculate the path, but cant you have both as a optino?
[00:04:37] <anonimasu> option?
[00:04:55] <fenn> sure
[00:05:05] <petev> I'm starting to like two modes here
[00:05:08] <fenn> just more complicated prorgamming
[00:05:13] <anonimasu> �hm yeah
[00:05:20] <petev> one is lermans non-optimal speed when path is important
[00:05:33] <petev> the other is SWPs FIR filter when speed is the issue
[00:05:38] <fenn> yeah
[00:05:43] <anonimasu> or maybe hm, could you have both of them at the same time?
[00:05:52] <fenn> anonimasu: mope
[00:06:08] <anonimasu> like a MXXX to flag a corner important
[00:06:09] <fenn> good, fast, cheap. pick two. emc's already cheap
[00:06:38] <anonimasu> although you can always slow down to get the accuracy even with the FIR filter..
[00:06:57] <petev> anonimasu: true under operator control or something
[00:07:07] <anonimasu> hm, cam programs do that with corners and stuff..
[00:07:25] <anonimasu> could you bound the max deviation the FIR filter can have?
[00:07:29] <anonimasu> *rants*
[00:07:44] <fenn> heh go read last night's log anon
[00:07:51] <petev> anonimasu: I think that would be path dependant
[00:08:10] <anonimasu> petev: operator control adds more human error to it..
[00:08:17] <anonimasu> logger_aj: bookmark
[00:08:17] <anonimasu> See http://solaris.cs.utt.ro/irc/irc.freenode.net:6667/emc/2005-12-09#T00-08-17
[00:08:45] <petev> ok, how about this
[00:08:57] <petev> we use lermans approach and make the look ahead adaptive
[00:09:12] <petev> just run full throttle until Q is full, etc.
[00:09:52] <petev> always adjust velocity to min of specified or obtainable from look ahead
[00:11:44] <lerman> That was the general idea. The look ahead would be such that we can guarantee a stop in time (so that we don't go off the road). But if the queue go too empty, we wouldn't be able to keep up with the RT component. So slow the RT down -- on the fly if necessary -- . Then we don't have to look that far ahead and can fill the queue again.
[00:12:35] <anonimasu> hm the trouble comes with tight segments with that..
[00:12:56] <anonimasu> like on a circle with multiple segments..
[00:13:07] <anonimasu> you should be able to go fast, but you cant because the queue is full..
[00:13:45] <petev> lerman: the RT processes segments at a constant rate
[00:13:52] <lerman> Are you sure? That should be easy to compute.
[00:13:57] <petev> TP output is positions with a time-law
[00:14:08] <k4ts> night all
[00:14:44] <petev> lerman: I guess I should clarify, RT after the TP
[00:14:51] <petev> the TP is in RT right now
[00:15:25] <lerman> petev. what I meant to say was that as the queue go empty, we would put in segments that have a lower speed. So each segment would be longer (in time).
[00:15:35] <petev> only the interpreter is user space
[00:15:46] <petev> lerman: which Q, from interpreter to TP?
[00:16:32] <lerman> petev: that would be the (future) connection in the new tp from the user space component to the RT component.
[00:16:55] <petev> in EMC2, the TP is already RT
[00:16:59] <anonimasu> night guys I'll read up on the previous discussion tomorrow
[00:17:03] <petev> gnight
[00:17:04] <anonimasu> and on the rest of this :)
[00:17:05] <lerman> The new tp could be partly in user space, I would think.
[00:17:12] <petev> right
[00:17:16] <anonimasu> kep up the good work :)
[00:17:18] <anonimasu> err keep..
[00:18:04] <lerman> gnight anon... No need to fix your typos -- we understand that your fingers are as bad as ours.
[00:20:42] <lerman> See: http://edgewise.pycs.net/weblog/2004/04/19.html%23P146 for: Voltaire said that "Best is the Enemy of Good" which apparently when translated from French into Russian and back to English via 1960's, Soviet Admiral S.G. Gorshkov, is "Perfect is the enemy of good enough."
[00:21:03] <petev> lerman: what's your summary of the sonja paper, I'm only about 1/4 through it and haven't gotten much out of it yet
[00:22:58] <lerman> I haven't read the whole thing. But my quick take is that it is for robots. Now that gets me back to Best is the Enemy of Good. Are we really going to design a 6 axis system with reverse kinematics for robots when most of use are using machines with three orthogonal axis?
[00:23:28] <petev> I'm happy with trivkins
[00:23:29] <lerman> My answer to the issue is that we need to fix up the framework so that we can plug in a variety of solutions.
[00:23:39] <petev> but smooth jerk limited motion would be nice
[00:24:25] <fenn> kinematics has nothing to do with trajectory planning
[00:24:37] <lerman> Then, each of us (who give a damn) could build his own replacement and we could play with them. Eventually, that might evolve to something that could be used for robots (and trivkins), but we wouldn't have to wait for it.
[00:24:55] <fenn> lerman: where did you get the idea that kinematics had anything to do with it?
[00:25:06] <fenn> we're talking about moving a point through space
[00:25:14] <petev> fenn: I'm not so sure, depends on whether stuff is being done task space or joint space
[00:25:16] <lerman> fenn: I think it does -- if you are limiting angular accelerations, but specifying trajectories in 3 space.
[00:25:22] <petev> there are some joint space algorithms
[00:26:42] <fenn> if you read the original post that started all of this.. what josh is suggesting is planning out the path in cartesian space, then applying whatever kinematic transform to get to joint space, and doing all the limiting in joint space
[00:26:54] <lerman> The reason to limit acceleration is because of physical limitations of the joints -- whether they are linear, rotational, or whatever.
[00:27:32] <petev> fenn: that's one approach, but we should consider everything
[00:29:00] <lerman> That would require that each joint do its own limiting and provide feedback to the feedrate setter -- whatever that is.
[00:29:16] <fenn> arrrrg!
[00:29:39] <lerman> I thought he was converting everything to b-splines in joint space.
[00:29:52] <fenn> yep, parameterized in terms of distance along the path
[00:30:21] <petev> fenn: let's take a simple case
[00:30:29] <fenn> for robots, you would take the worst-case accel and velocity limits, and apply that to all cases
[00:30:31] <petev> how about a circle in X, Y plane
[00:30:40] <lerman> That lets you set feedrate nicely.
[00:30:45] <petev> what planning limits do you use for the TP?
[00:30:56] <petev> individual joint limits or some global limits?
[00:31:01] <lerman> (parameterized in terms of distance)
[00:31:11] <fenn> petev: global limits is much simpler
[00:31:33] <petev> fenn: ok, then you do you gaurantee the global limits don't exceed a joint limit?
[00:31:43] <fenn> yes
[00:31:48] <lerman> But global limits don't do the job. The actuator force is what you need to limit.
[00:31:56] <fenn> if you wanted, you _could_ use the joint limits for a given point in space
[00:32:15] <fenn> actually now that i think about it it's not that hard
[00:32:20] <petev> I think it's not so easy
[00:32:44] <petev> if you set the global limits low, like in the circle case, to the accel one axis can provide
[00:32:52] <petev> it works, but maybe not optimally
[00:33:01] <fenn> right
[00:33:18] <fenn> it will be optimal for machines with trivial kinematics
[00:33:31] <lerman> Well, I think you've prooved my thesis that we need to worry more about the framework so that we can work with multiple approaches.
[00:33:35] <petev> fenn: not really
[00:34:11] <petev> lerman: what is wrong with the framework in the RT?
[00:34:22] <fenn> if the global limits are just a little more sophisticated than one variable
[00:34:47] <petev> fenn: yes, you will start to need limits for different planes, etc.
[00:34:48] <fenn> like a struct with limits for each cartesian axis
[00:34:59] <petev> u will be heading towards joint limits
[00:35:11] <lerman> I don't really know. But the discussion that the originators couldn't retrofit a new TP scares the hell out of me.
[00:35:26] <petev> lerman: RT is in very good shape with HAL
[00:35:28] <lerman> Or was I misreading the discussion?
[00:35:35] <petev> all above that is a mess IMHO
[00:35:53] <lerman> Ahhh. I'm please to hear that (it is good shape with hal)
[00:35:56] <petev> I have started to work on replacements for some of the higher level stuff
[00:36:18] <fenn> btw did you ever get any xml stuff written?
[00:36:30] <petev> just the sample config
[00:36:40] <petev> I spent time looking at packages and making the spread sheet
[00:36:57] <petev> did u try the config in kxmledit?
[00:37:05] <fenn> the one on the wiki?
[00:37:16] <petev> no, I emailed one last night
[00:37:34] <petev> more complete than the wiki version
[00:38:03] <petev> fenn: does your SF email work?
[00:38:10] <fenn> i think so
[00:38:20] <petev> did u get any emails from me?
[00:38:24] <petev> I sent 2
[00:39:05] <fenn> i see TP and Re: TP
[00:39:56] <petev> hmm, I sent one on XML too last night
[00:42:33] <lerman> Well, I'm out of here. Thanks for listening to my thoughts (or rants).
[00:43:13] <fenn> maybe i'll try to catch up on email before ranting some more
[00:43:18] <petev> ok
[00:54:23] <fenn> line 45 should that be tool id="35" ?
[00:54:45] <petev> hang on
[00:56:33] <SWPadnos> man - that was a lot to read up on :)
[00:56:38] <petev> yeah, that's odd, I saved it from kxmleditor last night and it was fine
[00:56:40] <SWPadnos> too bad anon and lerman left
[00:57:10] <petev> what did you think of using lermans approach with the adaptive Q length?
[00:57:22] <SWPadnos> I was going to suggest it, but he already had ;)
[00:57:52] <SWPadnos> you look ahead until either the minimum stopping distance, or the maximum lookahead has been reached
[00:58:05] <SWPadnos> if you hit max lookahead, then you need to slow down
[00:58:27] <petev> yeah, and limit velocity based on min of desired and that allowed by look ahead
[00:58:39] <SWPadnos> it should be possible to look ahead by 100 or more segments though, so it shouldn't be a limit in any but the most extreme cases
[00:58:50] <SWPadnos> yes
[00:58:54] <petev> I think max look ahead can just be limited by running out of bufs
[00:59:09] <SWPadnos> no - it's processing time that's the issue
[00:59:19] <SWPadnos> you may not be able to look at more than 64 or 128 segments in RT
[00:59:25] <petev> ahh, I'm talking about in the Q
[00:59:37] <SWPadnos> ok - queueing is a separate issue, IMO
[00:59:41] <petev> yes, in the TP u only look far enough to make sure u can stop
[00:59:56] <SWPadnos> did my email make sense to you?
[01:00:31] <petev> kinda, but I didn't see how a velocity smoothed path in joint space equated to limited jerk in task space
[01:00:42] <SWPadnos> ah - ok
[01:00:47] <petev> isn't it just vector addition for a normal machine?
[01:01:10] <SWPadnos> it should be, actually for normal or non-normal machines
[01:01:44] <petev> so if jerk is not limited in joint space, it won't be in task space either
[01:02:01] <SWPadnos> if the RT code gets poses in joint space, then a method that decomposes into joints, then does overall limiting after recombination, would work for any kinematics
[01:02:15] <petev> so u need to use a velocity smoothing algo like the cycloidal one that limits jerk
[01:02:37] <SWPadnos> could be - I don't know what that is ;)
[01:02:46] <petev> in the one paper I sent
[01:02:54] <SWPadnos> ok - haven't had a chance to read it yet
[01:02:59] <petev> it looked like the slickest smothing algo I have seen so far
[01:03:06] <SWPadnos> (invoicing, emailing, general accounting - yuk)
[01:03:11] <petev> yep
[01:03:36] <petev> I have had more time the last couple of days because I got my board out to fab
[01:03:49] <petev> it will get assembled next week, then I'll be back at it
[01:04:00] <SWPadnos> licky guy - I'm still trying to get my customers to provide specs ;)
[01:04:04] <SWPadnos> lucky
[01:04:21] <petev> hah, I have to write the specs for them
[01:04:25] <SWPadnos> but maybe I'll be able to afford Altium when they pay these invoices :)
[01:04:38] <SWPadnos> what PCB / schematic software do you use?
[01:04:49] <petev> I use OrCAD/PADS
[01:05:02] <SWPadnos> ok - what's the cost of those these days?
[01:05:12] <SWPadnos> I haven't used OrCAD since the SDT-III days
[01:05:19] <petev> not sure, have had them a while, I think a couple K
[01:05:53] <SWPadnos> do you know if there's a Linux version?
[01:06:29] <petev> no, all my CAD/CAM is on doze and I can't afford new version to switch to linux, so I don't even look
[01:06:56] <SWPadnos> yeah - that's my problem. Altium is $10k, but it's really slick
[01:07:00] <SWPadnos> and runs only on Win
[01:07:13] <petev> is that the embedde SW tools?
[01:07:26] <SWPadnos> schematic, PCB, FPGA, compilers, soft cores, etc.
[01:07:40] <petev> hmm, haven't seen that
[01:07:51] <SWPadnos> they own Tasking now, so you get Tasking C for all the soft cores
[01:07:54] <petev> I use synplicity and model sim for FPGA stuff
[01:08:05] <SWPadnos> check out the demos of the latest version
[01:08:38] <SWPadnos> the cool par tis that you can make a design for a Xilinx part, decide to use a Lattice part, and just retarget it (in about 5 clicks)
[01:09:23] <petev> what do you do your FPGA design in? I just use VHDL so re-target isn't too bad
[01:09:43] <SWPadnos> I don't do much yet, so I'd like a nice system to get me started ;)
[01:09:51] <SWPadnos> I have the Altera and Xilinx tools
[01:10:14] <petev> yeah, the FPGAs are so big now, the free tools handle parts big enough for most jobs
[01:10:19] <SWPadnos> yep
[01:10:41] <petev> you can do VHLS in altera/xilinx, but it's not that pretty
[01:10:45] <SWPadnos> I actually have the registered versions of both (with the inexpensive dev kits)
[01:10:46] <petev> they are a bit limited
[01:11:05] <SWPadnos> well - it looks like OrCad is also Windows only
[01:11:08] <SWPadnos> bastids
[01:11:10] <petev> I had registered versions way back, but the free ones do more than those now
[01:11:31] <SWPadnos> yeah - I got them for the NIOS and MicroBlaze soft cores
[01:11:36] <petev> probably since Cadence bought them and they don't want it to compete with their workstation SW
[01:11:56] <petev> have u been to www.opencores.org?
[01:12:00] <SWPadnos> at the embedded systems conferences, the kits are usually $200 - $500, including a nice dev board
[01:12:08] <SWPadnos> yep - just got the 7-CD set
[01:12:09] <petev> not always the cleanest code, but some interesting stuff
[01:12:25] <SWPadnos> annoyingly, most of the "open source" software is for Windows
[01:12:36] <petev> go figure
[01:12:52] <SWPadnos> I'm just about ready to give in and buy Windows XP 64 for my Opteron machine
[01:13:04] <petev> oh no
[01:13:05] <SWPadnos> but I'd be really annoyed to have to do that
[01:13:22] <petev> I won't touch anything after NT 2K
[01:13:36] <petev> the license on the new stuff is bad
[01:13:37] <SWPadnos> I can use VMWare, but it runs at like 80% speed max, and doesn't emulate multiple monitors
[01:14:20] <SWPadnos> yeah - but 2K is out of servicea s of middle of next year, so all the vendors will require XP soon - might as well bite the bullet at install time, rather than waiting to upgrade
[01:14:38] <SWPadnos> (out of service = no longer supported by MS)
[01:14:53] <petev> like they fixed all the bugs anyhow
[01:15:07] <SWPadnos> well - now they will "officially" not fix bugs
[01:15:43] <petev> ok, so on TP stuff
[01:15:50] <SWPadnos> ya
[01:15:51] <petev> if we smooth in the joint space
[01:16:14] <petev> how does the TP gaurantee that joint limits are respected when planning in the task space?
[01:16:38] <SWPadnos> thats a tough one - it has to use inverse kinematics (or the other way around)
[01:16:55] <petev> so how does the TP plan the path?
[01:17:06] <fenn> you can run the joint limits through inverse kinematics for each point on the path
[01:17:10] <SWPadnos> actually - why don't we see if it's possible to separate then combine joint limits?
[01:17:33] <SWPadnos> or computationally feasible
[01:17:40] <fenn> what does that mean?
[01:17:45] <SWPadnos> one sec
[01:17:46] <fenn> (havent finished reading email yet)
[01:17:56] <SWPadnos> ok
[01:18:04] <SWPadnos> just checked to see if it went to you ;)
[01:18:49] <petev> how does the TP handle arcs now?
[01:18:58] <petev> does it just sampled them at the TP rate?
[01:19:05] <petev> dam, I can't type
[01:19:11] <SWPadnos> actually - the machine setup should take into account the maximum feedrate possible at any part of a non-trivial machine
[01:19:33] <SWPadnos> so anything acceptable in the planner should be possible in the interpolator
[01:19:43] <petev> swp: did u get what I was saying about kinematics vs dynamics?
[01:19:49] <SWPadnos> yes
[01:19:52] <petev> ok
[01:20:06] <SWPadnos> and emc doesn't do dynamics - we have no way of controlling force on most machines
[01:20:16] <SWPadnos> or sensing it, for that matter
[01:20:26] <petev> I think dynamics is way too much computation to do real time
[01:20:30] <fenn> this is where orocos kinematics & dynamics library makes an entrance
[01:20:55] <petev> fenn: have u looked into them in detail?
[01:21:01] <fenn> no :(
[01:21:09] <SWPadnos> I'm not sure of that - there is no physical interface to a "force provider" or a "force sensor", so we can't use it in a control system
[01:21:24] <petev> swp: that's not needed
[01:21:38] <fenn> you know the mass of the elements (or can guess at them)
[01:21:42] <petev> dynamics just takes into account the forces required to do a given motion
[01:21:52] <petev> yes, u need things like mass, etc.
[01:21:56] <SWPadnos> it is, if you want to deal with force issues (ie, dynamic acceleration parameters)
[01:22:04] <SWPadnos> you need cutter load as well
[01:22:10] <petev> right
[01:22:19] <petev> but that can be calced
[01:22:21] <fenn> coefficient of modulus, dynamic bending moment
[01:22:25] <SWPadnos> and friction, which will be dynamic
[01:22:34] <petev> yep, a model of the machine
[01:22:37] <fenn> haha good luck
[01:22:39] <petev> that's why it's tough
[01:22:46] <SWPadnos> no it can't - it depends on how much material is actually being cut
[01:22:50] <petev> I don't even think we should consider it
[01:22:59] <SWPadnos> nope - let's drop it now ;)
[01:23:03] <petev> yep
[01:24:09] <petev> swp: did u read the sonja paper?
[01:24:10] <fenn> petev: what is the general gist of the .xls document?
[01:24:17] <SWPadnos> ok - so, is it possible to decompose a multiaxis move into separate single-axis moves, and then recombine them, in a "reasonable" amount of time, and if so, how is it done?
[01:24:24] <fenn> just that cutting forces > accel forces for steel?
[01:24:25] <SWPadnos> I read part of it, but it was a while ago
[01:24:47] <petev> fenn: the general gist is that we are doing a kinematics based TP
[01:25:07] <petev> therefore there are different limits for planning that the real machine limits
[01:25:19] <petev> the head room will depend on the type of machine
[01:25:34] <petev> for a BP, planning probably gets about 20% of the total power
[01:25:36] <SWPadnos> and as jmk pointed out, a router with a 1/8 bit will be just the opposite
[01:25:44] <petev> the rest is for cutting force
[01:25:59] <petev> swp: yes, on a gantry wood router it would be way different
[01:26:09] <SWPadnos> or a sign cutter, for example
[01:26:15] <SWPadnos> vinyl knife cutting
[01:26:32] <fenn> isn't that all done by the cam software and represented by the F word?
[01:26:54] <SWPadnos> that's my view - the path and speed are provided, so those are the target
[01:27:02] <petev> CAM doens't know about machine parameters
[01:27:07] <SWPadnos> yes it does!
[01:27:14] <SWPadnos> it needs to
[01:27:18] <petev> I mean like accel, etc.
[01:27:19] <fenn> yes it needs to
[01:27:20] <petev> no way
[01:27:32] <SWPadnos> OK - not necessarily accel
[01:27:44] <SWPadnos> though it should know about that
[01:27:54] <petev> most CAM you enter the cutter type and pick something off the feed/speed chart for material
[01:28:14] <fenn> which is based on a bridgeport
[01:28:28] <petev> fenn: what is based?
[01:28:34] <fenn> the feed/speed chart
[01:28:49] <SWPadnos> no - it's based on HP requirements for machining certain materials
[01:28:52] <petev> no, based on material, cutter type, cutter material, coolant, etc.
[01:28:59] <petev> yes, and HP
[01:29:42] <petev> the guy running the CAM needs to know the machine limits and not generate something that the machine can't hanlde
[01:30:08] <petev> in the packages I have used, there is nothing machine specific until the post
[01:31:00] <fenn> seems like the cam ought to be able to optimize it for the machine
[01:31:12] <fenn> a bridgeport can hog out more metal than a router can
[01:31:15] <SWPadnos> OK - I see a feedrate setting, but no accel (and no chipload either - how cheap)
[01:31:31] <petev> yeah, maybe in next generation sw or something, but now the guy running the CAM does the optimization
[01:31:41] <fenn> that's totally lame
[01:31:51] <petev> it's state of the art today
[01:31:52] <SWPadnos> a bridgeport with a 1HP spindle can't hog out as much as a Bridgeport with a 4HP spindle though
[01:31:59] <petev> I have VisualMill and MasterCAM
[01:32:07] <petev> both are about the same in capabilities
[01:32:15] <petev> MasterCAM has a little more AI
[01:32:16] <SWPadnos> of course, my software was $75
[01:32:19] <fenn> they are written by the same guy probably
[01:32:37] <fenn> and even he admits that they suck
[01:32:44] <petev> fenn: not VisualMill and MasterCAM
[01:33:51] <SWPadnos> petev, have you added yourself to Frappr?
[01:34:01] <petev> no, what;s that
[01:34:17] <SWPadnos> check the topic description :)
[01:34:24] <SWPadnos> EMC usage map ...
[01:34:43] <petev> oh, well I don't exactly use it yet unless on the bench counts
[01:34:54] <SWPadnos> I only use it on the bench as well ;)
[01:35:01] <SWPadnos> (one day, it'll be on the Bridgeport)
[01:35:08] <petev> yep
[01:36:02] <SWPadnos> you're on the west coast, right?
[01:36:28] <petev> yeah, CA
[01:36:39] <petev> ok, just added
[01:36:44] <SWPadnos> OK - do you get to the Embedded Systems conference in SF?
[01:36:57] <petev> very rarely
[01:37:05] <SWPadnos> well - it's only once a year ;)
[01:37:23] <petev> yeah, but it's been years since I went
[01:37:24] <SWPadnos> and you're right next door
[01:37:28] <petev> I know
[01:37:38] <petev> I hate going to the city
[01:37:41] <SWPadnos> heh
[01:37:44] <petev> no parking, etc...
[01:37:56] <SWPadnos> I love it (the House of Prime RIb is one of my favorite restaurants)
[01:39:13] <petev> what are the fastest speeds people have gotten out of the current TP with proper behavior?
[01:39:35] <SWPadnos> I think Les is the record holder, but I don't remember the speeds they got
[01:40:10] <petev> I was not impressed with my friends machine that I help him setup on EMC1
[01:40:34] <SWPadnos> what kind of hardware?
[01:40:37] <petev> I put 2.46 HP servos on it, and it was a dog no matter how we tuned it
[01:40:44] <SWPadnos> (both machining and computing)
[01:40:50] <petev> it's a little bigger thatn a BP, has box ways
[01:41:04] <SWPadnos> what kind of servos and drivers?
[01:41:09] <petev> I don't think there was a computing problem
[01:41:24] <petev> they were Emerson drives/motors, AC servo
[01:41:35] <petev> Vital motenc board
[01:42:06] <SWPadnos> hm - direct drive or reduction from motor to screw?
[01:42:08] <petev> I could get much better moves from the emerson drives with the PC control SW connected via RS232
[01:42:20] <petev> belt reduction to ball screw
[01:43:26] <petev> the only time it moved fast was when the homing algo blew up and it ran the X axis screw out of the nut ;-)
[01:43:37] <petev> that was scary
[01:43:44] <SWPadnos> heh - I'll bet
[01:44:17] <petev> but it did show what the machine was capable of
[01:44:49] <SWPadnos> well - one direction is very different from back and forth that most servos are asked for
[01:45:18] <petev> yeah, but it didn't even perform well in a straight line with EMC1
[01:45:53] <petev> I think there may be some serious bugs in the current TP
[01:45:54] <SWPadnos> there's no reason for that except bad ini settings (or bugs)
[01:46:01] <petev> yep
[01:46:09] <SWPadnos> no - you should be able to G0 and get top speed
[01:46:18] <SWPadnos> that has nothing to do with the planner
[01:46:22] <petev> I tried everything with PID and it didn't help
[01:46:51] <SWPadnos> hm - I'd be curious about it, but it probably doesn't matter at this point
[01:46:53] <petev> not sure if I recall what happens with G0, but jog was poor
[01:47:09] <petev> there was a huge lag in the jog control and it would move after you stopped
[01:47:36] <petev> yeah, I think the TP is going to be a re-write, so it doesn't really matter
[01:48:03] <SWPadnos> jog can be weird - in non-continuous mode, it treats it like a series of commands to move the jog increment
[01:48:16] <SWPadnos> so if the increment is 0.1 inch, it may move after you release the key
[01:48:23] <petev> it acted like they were getting Qed up or something
[01:48:26] <SWPadnos> (or 1 inch, which I also think is an option)
[01:48:38] <SWPadnos> that may be, due to keyboard repeat
[01:48:43] <fenn> i think that's how it works
[01:48:53] <fenn> each time a key is pressed a "jog" message is sent
[01:48:56] <petev> that's poor, it should stop when u let go
[01:49:05] <petev> the only feedback u have is visual
[01:49:13] <SWPadnos> that's what it does in "continuous" mode
[01:49:20] <fenn> i might be wrong
[01:49:40] <fenn> howdy fellas
[01:51:46] <petev> ok, ch 4 of sonja and I think all I have gotten is how to do a velocity blend on a single line segment between way points
[01:52:09] <SWPadnos> I'm just looking for pictures, then reading the surrounding discussion
[01:52:13] <fenn> if academics were straightforward they would be out of a job
[01:52:28] <petev> I should just skip to the summaries
[01:52:43] <fenn> that's why herman bruynix has so much trouble getting funding :)
[01:53:06] <SWPadnos> that's my technique. I usually don't understand the explanations and prrofs anyway - but I get the conclusions ;)
[01:53:11] <SWPadnos> proofs
[01:55:32] <petev> this may be stupid
[01:55:54] <SWPadnos> OK - the "non-jerk-limited" accel profile is called LSPB - "Linear Segments with Parabolic Blends"
[01:55:58] <SWPadnos> (page 14, biottom)
[01:56:09] <petev> but if you blend velocities in the joint space, won't you get a nice blend in task space since they are linear vector ops
[01:56:26] <SWPadnos> I would think so - that's why I want to do things on a joint basis
[01:56:39] <petev> why does she go throught the trouble to blend in task space then?
[01:56:47] <SWPadnos> dunno
[01:56:53] <petev> seems much harder than with only one dimension to deal with
[01:56:55] <fenn> kinematic transforms are not necessarily linear
[01:57:08] <SWPadnos> mmm - true
[01:57:12] <petev> ahh, but for trivkins we're good
[01:57:22] <SWPadnos> which means we're not good ;)
[01:57:34] <petev> I can't think about that robot stuff, I just want my BP to work well
[01:57:41] <SWPadnos> there are a few things that would be great to ignore, but they can't be:
[01:57:51] <fenn> emc2 already works with bridgeports
[01:58:00] <SWPadnos> 1) non-trivial kinematics are a requirement
[01:58:16] <SWPadnos> 2) operation on non-bleeding-edge computers is a requirement
[01:58:43] <skunkworks> Me again. ready for another paragraph? Adding the extra variable to the emc.ini and the core_stepper.hal for head room allowed me to accelerate the slow axis (z) faster. thanks again. I was able to accellerate it at 4ips^2. I had toset the headroom to the core_stepper hal to almost twice the acceleration around 8ips^2. I still need to be able to set the other 2 axises to a higher acceleration. they can do over 10ips^2. there does
[01:58:45] <SWPadnos> 3) the ability to have remote GUI is a requirement (but can safely be ignored in the TP)
[01:59:29] <SWPadnos> (I'm sure there are more)
[02:00:00] <petev> can we limit the kinematics classes we need to deal with?
[02:00:02] <SWPadnos> that doesn't mean that a 10-year old PC has to be able to control a 2000 IPM router though
[02:00:36] <SWPadnos> as long as we include trivial, stewart platforms, and PUMA-style, yes ;)
[02:01:01] <SWPadnos> I don't think that it would be good to make a new EMC that can't do what the old emc does (in the user's eyes)
[02:01:01] <petev> ok, I'm not falimiar with those, are the fwd/rev kins linear?
[02:01:06] <SWPadnos> no
[02:01:13] <SWPadnos> a stewart platform is a hexapod
[02:01:25] <petev> darn
[02:01:47] <fenn> why are you trying to blend in task space?
[02:01:58] <fenn> to keep errors in tolerance?
[02:02:04] <petev> I'm not, I'm wondering why sonja is
[02:02:23] <SWPadnos> I don't think we should worry about where the blending is done
[02:02:29] <petev> I'm asking how a blend in joint space will translate to task space
[02:02:37] <SWPadnos> if we come up with an algorithm that can be run in RT, then it will be
[02:02:41] <fenn> it will be vrey close for small blends
[02:02:42] <petev> it's easy to see for trivkins
[02:02:52] <petev> not clear on others
[02:02:58] <SWPadnos> wait - joint <-> cartesian (or world), not task
[02:03:10] <SWPadnos> task vs RT, joint vs. world
[02:03:11] <fenn> whats the diff between task and cartesian?
[02:03:17] <fenn> oh
[02:03:18] <petev> task is cartesian or world
[02:03:35] <SWPadnos> just limiting the probability of confusion
[02:03:38] <fenn> task space is more of a robotics concept
[02:03:43] <SWPadnos> since task also means "the task controller"
[02:03:45] <fenn> "pick up an object here"
[02:03:56] <SWPadnos> sure - it would be the work envelope
[02:04:15] <fenn> but which basis vectors?
[02:04:53] <fenn> i would say cartesian
[02:04:54] <SWPadnos> click on the image here for a nice stewart platform: http://www.i-way.co.uk/~storrs/lme/LMEHexapodMachine.html
[02:05:07] <fenn> ignore the frame design though - ick
[02:05:57] <fenn> here's my crappy hexapod summary page: http://www1.atwiki.com/gingery_machines/pages/hexapod
[02:06:05] <petev> ok, it's like 3 joints, but uses 6 for tilt?
[02:06:27] <SWPadnos> no - it gves a full 5DOF (possibly 6-ish), and uses 6 actuators
[02:06:39] <SWPadnos> you need all 6 actuators though
[02:06:49] <petev> right, but without the head tilt only 3 joints are needed?
[02:07:03] <SWPadnos> no - you need all 6 all the time, I think
[02:07:09] <fenn> if you have 3 joints its not a hexapod
[02:07:13] <petev> I don't get it then
[02:07:15] <fenn> called a "deltabot" or something
[02:07:24] <petev> fenn: sure, just trying to understand how it moves
[02:07:31] <fenn> a deltabot is not as rigid as a hexapod
[02:07:42] <SWPadnos> think of the string around a drum
[02:07:53] <petev> seems like the joint pairs would move the same without tilt
[02:08:02] <SWPadnos> you get three triangles pointing down, and three triangles pointing up
[02:08:22] <SWPadnos> you can vary the lengths of two sides of each triangle (the 6 joints are the shared edges)
[02:08:32] <SWPadnos> I need to run for a bit - bbl
[02:08:47] <fenn> great now i have to explain hexapods again
[02:09:02] <petev> fenn: if the head did not tilt, couldn't you do it with 3 joints?
[02:10:10] <fenn> you would need to use 3 parallelograms
[02:11:10] <fenn> or you can use a variable length boom, like this http://www.abb.com/Global/brabb/brabb155.nsf/viewunid/DCF16ACD76EF8A5A03256D10005FA10B/$file/IRB+940+Tricept.jpg
[02:11:15] <petev> in the pic in the first line, if u replaced each joint pair with 1 joint, seems like u could still control the head, but tilt would be limited
[02:11:27] <petev> line/link
[02:11:30] <fenn> no it would just flop all over
[02:14:10] <petev> I think I see what u are saying, but only because the 3 joints are not connected at a single point on the head
[02:14:40] <fenn> this is a deltabot picture: http://www.industrial-robots.com/Images/parallel/spider.jpg
[02:15:30] <fenn> used for pick and place, so rigidity is not important
[02:17:40] <petev> ok, but if I took 3 joints and connected them at a point, it would be like a 3 sided pyramid, you just wouldn't have the control to angle the head, but you could make that single point trace paths out in 3D space it seems
[02:18:01] <fenn> yeah
[02:18:23] <petev> ok, that's what I was trying to say so poorly
[02:19:17] <fenn> the problem is that the head tilts, but you cant do anything about it
[02:19:23] <petev> right
[02:19:36] <petev> I was just trying to understand the basic motion
[02:20:15] <Jacky^> G night all
[02:20:17] <fenn> there are some movies around but i cant find them
[02:20:26] <Jacky^> Jacky^ is now known as Jacky^afk
[02:20:26] <petev> that would be cool
[02:22:34] <fenn> go to fenn.dyndns.org/pub/emc/screenshots/
[02:24:35] <petev> downloading... what did you think of the xml config?
[02:27:05] <fenn> um, it works..
[02:27:22] <petev> would u change it?
[02:27:32] <petev> what's this, a bugbot?
[02:28:03] <fenn> oops that might have gotten m ixed in with the hexapod movies
[02:28:09] <fenn> i hate those things
[02:28:30] <petev> ok, I'll try another
[02:29:29] <petev> I'm open to YAML or SLiP, but not debs and no good C/C++ libs
[02:29:38] <petev> I think this will be a problem
[02:29:46] <fenn> hexmov2 is a good one
[02:29:51] <petev> ok
[02:30:34] <fenn> the alex_joni_device is what you were describing
[02:30:46] <fenn> he calls it a tripod
[02:31:12] <petev> what does it do?
[02:31:17] <fenn> nothing
[02:31:29] <petev> just a test toy?
[02:31:32] <fenn> positions a washer in space
[02:31:36] <fenn> yeah just a toy
[02:31:38] <petev> ahh
[02:32:17] <petev> what format is mov?
[02:32:22] <fenn> quicktime
[02:32:23] <petev> I don't have a viewer
[02:32:26] <petev> oh
[02:32:28] <fenn> poop
[02:32:47] <petev> I'll save them and check it out later
[02:33:11] <fenn> hexvan(tage) is a crappy design.. i should delete that movie
[02:33:57] <petev> ahh, just saw the movie of the aj device
[02:37:01] <fenn> so, once you have a file in xml, parsing is easy?
[02:37:29] <petev> yes, there are libs that will parse it and make C data structs
[02:37:42] <petev> there are packages for python, lisp, etc. too
[02:45:10] <fenn> you guys assumed a certain mass for the moving parts of the machine tool for your spreadsheet?
[02:45:29] <petev> yeah, I have it as an input in red
[02:45:36] <petev> it was set at 600 lbs
[02:45:49] <petev> which is worse case with work on my BP
[02:45:54] <fenn> yeah
[02:46:25] <fenn> but a hsm won't have near that high of a moving mass
[02:46:37] <petev> hmm, not sure about that
[02:46:44] <petev> I'm not familiar with them
[02:46:53] <petev> change the red inputs and see what happens
[02:47:51] <fenn> guess i'll have to download 50 gnome packages after all
[02:47:59] <fenn> (dont have any spreadsheet software at all)
[02:47:59] <petev> for what?
[02:48:18] <petev> dam, SlickEdit is slick
[02:48:37] <petev> I opened the xml file and it understands it and flagged the errors
[02:49:23] <petev> u don't have any MS stuff?
[02:49:59] <fenn> hell no
[02:50:03] <petev> isn't there some office compatible apps for KDE?
[02:50:23] <fenn> only spreadsheet software i know of is gnumeric and star office
[02:50:40] <petev> is star office gnome or kde?
[02:51:15] <fenn> neither
[02:51:39] <fenn> what's the travel on you bp?
[02:51:57] <petev> x=18, y=12, z=6
[02:58:25] <fenn> wow the hexapode moving mass is 300kg
[02:58:33] <fenn> hexapode is a brand
[02:58:55] <petev> that's pretty high
[02:59:13] <petev> is it a big machine?
[03:01:44] <skunkworks> axis + lerman interp = pretty damn cool http://www.electronicsam.com/images/KandT/AXIS.JPG
[03:02:17] <cradek> what are the green droppings?
[03:02:35] <cradek> skunkworks: what's funny is I tried to rotate the view
[03:03:19] <skunkworks> funny - the mouse pointer - it must have been whiping when I captured it
[03:04:15] <cradek> skunkworks: we're this close --><-- to calling axis 1.1rc2
[03:04:39] <cradek> skunkworks: we're really happy with it right now. Any feedback welcome (not about new features we should add, but about bugs and usability)
[03:12:55] <fenn> petev: yes the hexapode is pretty big.. 2000 ipm rapids and 40KW spindle
[03:13:31] <skunkworks> can you change the 3d view perspective? the only comment I have so far is that the cone should stay the same size on the screen - or relitively - If I have a big file size wise and zoom in the cone gets huge. unless that has changed - just downloaded the latest version.
[03:14:25] <fenn> petev: it is comparing them to conventional machines with moving masses of 10-20 metric tons
[03:14:38] <fenn> dunno if thats a fair comparison though
[03:15:53] <skunkworks> I have not had a problem with axis as far as I have used it. great job
[03:16:17] <cradek> that's great
[03:16:19] <fenn> you can rotate with the middle button
[03:16:37] <cradek> fenn: not on the picture in my web browser
[03:18:22] <skunkworks> middle mouse button?
[03:18:40] <cradek> skunkworks: ?
[03:18:45] <fenn> usually the middle button is emulated by holding down both left and right buttons
[03:18:50] <fenn> or you press the scroll wheel
[03:19:03] <petev> scroll zoom is nice too
[03:19:19] <cradek> skunkworks: you did know you can rotate the view, right?
[03:19:53] <skunkworks> I know I can slide it around with the mouse - didn't know how to rotate the image. damn
[03:20:08] <skunkworks> keeps getting better
[03:20:09] <cradek> dammit man! read the help screen!
[03:20:17] <skunkworks> help?
[03:20:26] <fenn> i think rotate should be right button, zoom middle button
[03:20:28] <petev> cradek, it was so intuitive, no help screen needed
[03:20:30] <cradek> did you know you can click a line on the screen and it jumps to the corresponding line in the program?
[03:20:34] <petev> it's like most CAD programs
[03:20:48] <cradek> petev: thanks, that was among our goals, but I guess not everyone tried all the buttons
[03:20:57] <skunkworks> I know the click thing
[03:20:59] <fenn> not everyone has a 3 button mouse either
[03:21:01] <skunkworks> thanks ;)
[03:21:04] <petev> yeah, and fenn doesn't use doze programs
[03:21:07] <skunkworks> I don't
[03:21:13] <skunkworks> have a 3 button mouse
[03:21:33] <fenn> if that's the standard interface then dont mess with it
[03:21:37] <cradek> I thought we all had three buttons since about 1995
[03:21:49] <cradek> skunkworks: the help/quick reference screen
[03:21:54] <fenn> only linux dorks have 3 button mice
[03:22:12] <cradek> I have a four-button tablet - what does that make me?
[03:22:17] <fenn> an uber-dork
[03:22:25] <skunkworks> Oh well - tomarrow I will play with it some more and maybe figure out the acceleration problem in emc2
[03:22:30] <cradek> I'll take that as a compliment
[03:22:36] <skunkworks> thanks again and good night
[03:22:39] <cradek> goodnight
[03:22:48] <fenn> i wish i had a 4 button tablet.. sigh
[03:23:20] <cradek> I rigged mine to give scroll events if you hold button4 and move up or down
[03:23:30] <cradek> I really like it
[03:23:46] <fenn> i want a hexapod-shaped 3d mouse
[03:23:52] <petev> cradek, do you use that on your machine?
[03:24:02] <cradek> petev: sometimes
[03:24:14] <cradek> petev: I have one at work all the time
[03:24:25] <cradek> petev: I switched to a mouse on my milling machine's computer because of desk space limitations
[03:24:36] <petev> I got one of those 8" touch panel monitors for my machine
[03:24:47] <petev> it seems pretty good so far
[03:25:00] <cradek> petev: haven't seen those.
[03:25:05] <fenn> pete were you at the fest in illinois?
[03:25:12] <petev> fenn: no
[03:25:22] <petev> cradek: lillyput and all the others
[03:25:35] <petev> mostly intended for auto use I think
[03:25:42] <fenn> ok, was wondering if you were that guy i never talked to with the 8" touch panel
[03:25:50] <petev> nope
[03:26:00] <cradek> if skunky thought axis was super cool without knowing about the navigation, he's going to be impressed
[03:26:00] <petev> just set it up a few weeks ago
[03:26:07] <petev> yep
[03:26:26] <cradek> I wonder if it's a bug that he hadn't figured that out
[03:26:43] <cradek> it's pretty clear in the Quick Reference
[03:26:50] <cradek> but we don't want people to have to read that
[03:27:13] <petev> he probably just hasn't used many CAD programs
[03:27:25] <fenn> i havent used any cad programs and i got it
[03:27:46] <fenn> oh well
[03:27:49] <cradek> hmm
[03:27:54] <fenn> maybe a tooltip on the P button
[03:28:16] <cradek> maybe
[03:28:25] <cradek> some apps teach about obscure features with those startup tips
[03:28:34] <cradek> but I hate them with a passion
[03:29:12] <cradek> I think everything should be self-documenting and you should see how it works by looking at the screen and using your experience with how the widgets work
[03:29:18] <cradek> but that's a really hard thing to do perfectly
[03:29:30] <fenn> hard to document a 3d viewing area
[03:29:38] <cradek> yeah.
[03:29:55] <cradek> scrollbars would "document" translation but otherwise be pretty useless
[03:30:02] <cradek> I think you can't document the rotation
[03:30:19] <fenn> have a paperclip show up and start harassing you
[03:30:22] <cradek> why we put the summary in help for this very reason
[03:30:25] <cradek> s/why//
[03:30:37] <cradek> because we couldn't figure out a way to document the navigation on-screen
[03:34:30] <jepler> did skunkworks say whether he tried anything in the hopes that it rotated? Like a modified left mouse button or something?
[03:34:41] <cradek> he didn't say
[03:34:51] <cradek> I think it never crossed his mind that there was anything but the preset views
[03:35:01] <cradek> maybe the toolbar with the radiobutton-icons does give that impression
[03:35:27] <jepler> I wonder if the scroll-in-out button should be the wheel/middle button, and the rotation should be the right button
[03:35:39] <jepler> since the middle part of the mouse is already for scrolling
[03:35:50] <fenn> that's what i thought originally, but it's nice to be able to zoom and rotate at the same time
[03:36:16] <cradek> I don't think there's any standard (or consensus)
[03:36:23] <cradek> but I haven't used many of this kind of program
[03:36:26] <jepler> me either
[03:36:36] <jepler> the other one I know about puts a context menu on RMB so that wouldn't work
[03:36:52] <cradek> yeah I bet that's common
[03:37:21] <fenn> i dont think there's any standard either
[03:37:34] <fenn> it ought to be configurable
[03:37:42] <jepler> we mostly adopted the bindings that whoever designed Togl, the opengl widget for Tk, chose
[03:37:54] <jepler> or maybe whoever created the Python interface to Togl, I'm not sure
[03:38:16] <petev> I think it's fine the way it is
[03:38:49] <cradek> fenn: I think we both eschew configurability for the sake of configurability
[03:38:50] <petev> cradek: BTW, how do you get the data for the 3D plot?
[03:39:07] <cradek> petev: in general, we ask the emc interpreter for it
[03:39:21] <cradek> petev: that way it's sure to match up when the machine actually runs
[03:39:22] <petev> so you do reads, then ask for pos?
[03:39:22] <jepler> I'm sad that skunkworks didn't discover it on his own, and I'd like to find some way to make it discoverable.
[03:39:27] <jepler> startup tips are not an option, though
[03:39:39] <cradek> jepler: I agree - those suck
[03:39:41] <jepler> I'm with cradek on that one, and the issue of configurability
[03:40:26] <fenn> brlcad for example uses alt- for rotation, but that doesnt work with my window manager
[03:40:54] <cradek> petev: we tell the interpreter to canonify the file, and convert that canonical representation (lines, arcs) into GL display lists
[03:40:55] <fenn> alt- is the override so you can drag windows around when they are larger than the screen
[03:41:26] <petev> I dont't see that in the API, just read and execute with status info for pos, etc.
[03:41:32] <cradek> fenn: yeah, it's asking for trouble to assume modifiers other than control are free
[03:41:58] <jepler> petev: it's not in the API exactly; we embed a copy of the rs274ngc interpreter in the Axis program
[03:42:04] <jepler> petev: that's the purpose of the 'gcode' module
[03:42:05] <petev> ahh
[03:42:09] <petev> we need to fix that
[03:42:33] <jepler> fix it how?
[03:42:41] <petev> hopefully I can get this new interp done and have a proper API
[03:42:50] <jepler> the API is just fine
[03:43:02] <jepler> don't break the API, or I'll send you a bill for the time it takes to fix AXIS
[03:43:12] <petev> no, I think it leaves things to be desired, like the whole param mess
[03:43:57] <jepler> I guess I don't know the mess you mean
[03:44:15] <petev> right now you can't do params through the API
[03:44:31] <petev> axis is looking at the param file directly
[03:44:33] <jepler> Do you mean stuff like "set the value of #3" "fetch the value of #3"?
[03:44:36] <petev> hence all the sync stuff
[03:44:40] <petev> yes
[03:45:05] <petev> there are other re-start problems I want to fix too
[03:45:41] <jepler> def get_external_parameter_filename(self): # XXX not used and wrong
[03:45:41] <jepler> return "/usr/src/emc/sim.var"
[03:45:47] <jepler> this is the only place that AXIS refers to .var files
[03:46:05] <petev> do u have your own API on the axis end that wraps all the NML stuff?
[03:46:21] <jepler> petev: yes, that's the 'emc' module
[03:46:39] <petev> I don't follow?
[03:46:50] <petev> is that a python module?
[03:46:57] <jepler> Yes, 'emc' is a Python module installed by AXIS
[03:46:58] <petev> I'm not a pything guy
[03:47:02] <petev> ok
[03:47:18] <petev> so worst case a little tweaking there should take care of things
[03:47:27] <petev> that's good
[03:47:51] <jepler> yeah, every time you guys break the API we have to change AXIS a little bit
[03:48:14] <petev> well, we should fix the API first and document it
[03:48:18] <petev> it's broken now
[03:48:24] <cradek> and it's currently compatible with three different emc versions with (sometimes gratuitously) different APIs
[03:48:36] <jepler> Yeah, the #ifdefs are getting worse though
[03:48:58] <petev> so what would u suggest?
[03:49:11] <petev> I don't think the API allows all the operations it should
[03:49:37] <cradek> it might be nice to be able to read the variables
[03:49:43] <cradek> but I don't see any reason to allow writing to them directly
[03:49:44] <petev> that's for sure
[03:49:59] <jepler> petev: I don't mind real improvements
[03:49:59] <petev> and I think a lot of the canonical status/state should be available too
[03:50:12] <petev> jepler: I started from scratch
[03:50:22] <petev> so whatever you want you can have
[03:50:22] <cradek> do you mean like "we're currently running a G0/G1/G2"
[03:50:26] <petev> now's the time
[03:50:39] <jepler> petev: but some of the things that have really peeved me are the pointless changes from rs274ngc_foo to interp_foo (or the other way around)
[03:50:44] <petev> I see interp status/state and canon status/state
[03:51:02] <petev> jepler, none of that
[03:51:03] <jepler> or changing the return value of an api from 0/-1 to true/false to indicate success/failure
[03:51:19] <cradek> or changing capitalization of variable names
[03:51:20] <jepler> so I guess my criteria for a good change is that it actually improves something, instead of merely making it different
[03:51:30] <petev> the motivations were to fix real issues with status, re-start, code base size, etc.
[03:51:57] <petev> the code is much smaller, faster, more readable, and hopefully will fix these problems too
[03:52:32] <petev> It might not be well received by some as alot of work has been done adding to the interp lately, but I don't know what to do about that
[03:52:46] <cradek> petev: we seem to have a real problem keeping the units straight. It would be nice to step back and figure out how best to handle user units.
[03:53:03] <petev> in the new design, I will take care of that
[03:53:12] <petev> I already put it in the config file
[03:53:24] <petev> the current code is pretty hard to follow
[03:53:26] <cradek> well there's all sorts of stuff about units in the config file
[03:53:33] <petev> I'm not sure I want to attempt fixing it
[03:53:36] <cradek> the interp screws with the units, the gui screws with the units, and so on
[03:53:43] <petev> yep
[03:53:58] <cradek> screwing with the units is required somewhere, but hopefully only one place
[03:54:00] <petev> I don't want to convert the internal array because that is error prone
[03:54:07] <petev> both for coding and for rounding
[03:54:12] <cradek> the internal array should be in user units
[03:54:22] <petev> yes, and stay that way
[03:54:29] <cradek> I agree
[03:54:40] <petev> I added a new interp units to the xml I sent you
[03:54:59] <cradek> I did get your mail, but looking at xml causes me pain
[03:55:00] <petev> that way the user units can match lead screw and interp can still be what you want
[03:55:10] <petev> use kxmleditor
[03:55:11] <cradek> and I do not have kde (or debain) to install the viewer you were talking about
[03:55:22] <petev> I don't want to look at the raw xml either
[03:55:30] <petev> hmm
[03:55:32] <jepler> night all
[03:55:36] <cradek> g'night
[03:55:36] <petev> gnight
[03:55:41] <cradek> I should go home too
[03:55:49] <jepler> cradek: you're still at work? go home!
[03:55:51] <cradek> I came to work to fix some stuff up before morning
[03:55:55] <petev> is there no xml viewer for the linux u are running?
[03:55:56] <cradek> jepler: yeah yeah
[03:56:03] <cradek> petev: no idea, never needed one
[03:56:24] <cradek> goodnight all
[03:56:30] <petev> some browsers will handle it, but I'm not sure on linux
[03:56:39] <cradek> petev: we'll talk later, I'm interested in your efforts
[03:56:44] <petev> ok
[03:58:36] <fenn> heh konqueror attempts to display the xml by stripping out all of the tags
[03:58:47] <petev> oh, that sucks
[03:59:11] <petev> I think it will do the right thing with kxmleditor installed
[03:59:19] <petev> they are supposed to integrate
[03:59:26] <fenn> nope
[03:59:44] <petev> hmm, bad info in the docs, big surprise
[04:00:45] <fenn> konqueror is not good with file types
[04:00:59] <fenn> i think it relies on mime types which dont exist on a local filesystem
[04:01:08] <petev> hmm
[04:01:13] <petev> that could be it
[04:19:56] <CIA-12> 03jmkasunich * 10emc2/src/hal/utils/ (6 files): Improvements to halscope: it now saves all of its config on shutdown and restores it on startup; vertical offset entry works better; removed some unused code; general cleanup
[05:37:11] <CIA-12> 03paul_c * 10emc2-auto/wiki/ (18 files in 10 dirs): "Auto update wiki from a cron job. Fri Dec 9 05:30:02 GMT 2005 "
[07:31:39] <Jymmm> hola
[07:31:44] <anonimasu> 3hello
[07:32:14] <Jymmm> anonimasu you dont do wood do you?
[07:32:26] <anonimasu> nope
[07:32:31] <Jymmm> bastard!
[07:32:34] <anonimasu> heh
[07:32:35] <Jymmm> lol
[07:32:39] <anonimasu> you cant weld wood.
[07:33:01] <Jymmm> sure you can, just have to wait till it's petrified
[07:33:45] <anonimasu> heh
[07:33:52] <Jymmm> Just needed a comparison of dust on rails and lubing up.
[07:34:34] <Jymmm> anonimasu you running any kind of gantry machine?
[07:34:53] <anonimasu> heh
[07:34:54] <anonimasu> no
[07:35:20] <anonimasu> let me get a pic
[07:35:20] <Jymmm> you are such a lamer! wha da hell good are ya?!
[07:36:44] <Jymmm> found it/one yet?
[07:36:57] <anonimasu> nope
[07:37:07] <Jymmm> k
[07:38:35] <Jymmm> I'm finding my machine will loose steps, but if I relube it it's fine. Seems to be in one axis (X), just not sure if dust will cause this in less than 30 days
[07:38:41] <Jymmm> between lubes
[07:39:01] <Jymmm> or if it could be somethign else.
[07:39:39] <anonimasu> IN A SEC..
[07:39:50] <anonimasu> hmm oj
[07:40:10] <anonimasu> I cant find pics
[07:40:12] <anonimasu> wait a bit
[07:40:13] <Jymmm> I did notice that some of the plastic caps that cover the bolts are pushed in (not flush) and collecting dust, which might be getting trapped
[07:40:25] <Jymmm> on the rails themselves.
[07:40:35] <Jymmm> no worries on the pic.
[07:42:00] <anonimasu> got one now
[07:42:05] <anonimasu> not of the whole machine though
[07:42:57] <Jymmm> lets see it
[07:43:11] <anonimasu> http://www.bojn.net/~an0n/machinetest.jpg
[07:43:36] <anonimasu> it's a manual machine yet
[07:43:44] <anonimasu> I've got another one that's going to be cnc:ed
[07:43:49] <anonimasu> schaublin sv13
[07:44:05] <Jymmm> ok, so a mill then.
[07:44:08] <anonimasu> yeah
[07:44:18] <Jymmm> you could have just said that you know
[07:44:46] <anonimasu> ah well,
[07:44:51] <anonimasu> brb going to go to work :)
[07:44:52] <Jymmm> lol
[07:48:22] <Jymmm> Jymmm is now known as Red70sShow
[07:48:22] <Red70sShow> Red70sShow is now known as Jymmm
[07:49:03] <Jymmm> laters
[09:22:24] <anonimasu> hello alex_joni_
[12:57:30] <Jacky^afk> Jacky^afk is now known as Jacky^
[12:57:40] <Jacky^> morrrrning
[12:58:08] <chinamill> Good afternoon
[12:58:16] <Jacky^> hi chinamill :)
[12:58:52] <Jacky^> we got a troll on frappr ?
[12:59:20] <Jacky^> bauahha ! who is Thecooltool ?
[12:59:50] <chinamill> trolls are scary, right.
[12:59:52] <Jacky^> lol
[13:01:32] <alex_joni_> not a troll
[13:01:48] <alex_joni_> it's a company selling products..
[13:01:51] <Jacky^> hi alex_joni_ good ;)
[13:03:40] <chinamill> Do You have any clue about the relation between ferror and min_ferror in emc2? I had to put the ferror=12 and min_ferror=10 to be able to run without following errors, would this be normal? (metric setup)
[13:07:22] <Jacky^> chinamill: Ive no clue, I guess the choice of a value can depend on many others there around too
[13:08:11] <Jacky^> I just played with a lot of random value to get it working
[13:08:22] <chinamill> ok
[13:12:34] <chinamill> Do you know anyone running emc2 with metric style ini?
[13:15:06] <skunkworks> are you running steppers? Or servos? (I am running inch sorry)
[13:16:05] <Jacky^> to be honest I had some issue running with mm units in emc and large scale but it was a bug, probably fixed in emc2
[13:16:38] <Jacky^> Im using inches in emc2 too, and G21 command in my gcode files
[13:17:59] <skunkworks> right - that is what the penguin does and runs fine for me. (don't do much in the metric realm yet)
[13:18:45] <skunkworks> alex - saw some videos of your flying washer - pretty neet. was that using emc?
[13:19:08] <skunkworks> I think that was yours anyway
[13:19:51] <alex_joni_> emc2
[13:20:50] <skunkworks> it was funny - when I first looked at it I didn't see the cables. thought some how you where using 3 big electro magnets to run the washer around ;)
[13:21:09] <alex_joni_> that would be even cooler ;)
[13:21:15] <alex_joni_> but not precise enough I guess
[13:22:49] <skunkworks> Question - how did you do that in emc - to tell it that the 3 points where controlling the posision of the washer - looked like you where making strait lines and circles with it.
[13:22:58] <skunkworks> is it easyier that I am thinking?
[13:23:23] <alex_joni_> it's kinematics
[13:23:31] <alex_joni_> to convert from joints to carthesian
[13:23:57] <alex_joni_> usually emc is running with trivial kins (j0=x, j1=y, j2=z, ...)
[13:24:19] <alex_joni_> but you can define X=sqrt(j0*j0+j1*j1) or any formula to suit your machine
[13:24:59] <skunkworks> same concept as an arm (being able to convert the mutible joints into an end desired cordiate system
[13:25:13] <skunkworks> It may be a little over my head
[13:25:45] <alex_joni_> on RRR robots, usually coordinate system depends on most axes
[13:26:10] <alex_joni_> so X=f(j0..j5), Y=g(j0..j5), Z=h(j0..j5), etc
[13:26:25] <skunkworks> do you have a link to some info on this?
[13:26:52] <skunkworks> that is pretty neet
[13:27:15] <skunkworks> kinda makes sense
[13:28:20] <alex_joni_> read on google about kinematics
[13:28:25] <alex_joni_> forward and inverse kinematics
[13:28:36] <alex_joni_> after that look at pumakins.c in emc1 source code
[13:28:37] <skunkworks> thanks
[13:29:18] <alex_joni_> http://cvs.sourceforge.net/viewcvs.py/emc/emc/src/emcmot/puma560kins.c?rev=1.3&view=markup
[13:30:18] <alex_joni_> a bit of math though :D
[13:30:58] <skunkworks> I did well in "math" but that was 15 years ago. thanks again
[13:31:07] <alex_joni_> np
[13:31:48] <skunkworks> who do you think would be able to help me with the acceleration problem? wanting to run 2 axises at a faster acceleration than the 3rd?
[13:32:04] <alex_joni_> hang on till this weekend
[13:32:07] <alex_joni_> or after the weekend
[13:32:13] <alex_joni_> * alex_joni_ and cradek will be working on it
[13:32:19] <alex_joni_> we have a plan ;)
[13:32:30] <skunkworks> I will be around - thanks.
[13:32:30] <alex_joni_> to rule the world.. but don't tell anyone
[13:32:34] <alex_joni_> or I'll have to kill you
[13:32:34] <alex_joni_> :D
[13:32:38] <skunkworks> funny
[13:32:39] <anonimasu> alex_joni_: dont steal my plan
[13:32:50] <alex_joni_> you can join
[13:32:59] <anonimasu> hm sounds nice
[13:33:00] <anonimasu> :D
[13:33:01] <anonimasu> I am in
[13:33:26] <skunkworks> did you see this? I got the latest emc2 (lerman interp) and axis. http://www.electronicsam.com/images/KandT/AXIS.JPG
[13:33:29] <skunkworks> I am in too
[13:34:00] <skunkworks> I guess my math fundimentals are still there anyway (geometry) ;)
[13:34:10] <alex_joni_> that looks nice
[13:34:43] <skunkworks> I like do-while loops ;)
[13:34:55] <alex_joni_> what are those green thingies?
[13:35:39] <skunkworks> slow computer - the mouse was whiping when the capture was made - I guess I will have to make a better one - you are the second person to ask ;)
[13:42:05] <skunkworks> I was accelerating z at 4ips^2 last night - but found out it was loosing steps. So I slowed it down - still loosing steps - change the scaling to 2000 vs 20000 - still lossing. slowed it way down - still loosing. pretty sure it isn't emc2 - the other 2 axises will reposition perfectly - after many many cycles. thinking it is a stepper driver problem or maybe a cable
[13:47:00] <alex_joni_> sounds like that
[13:47:07] <alex_joni_> alex_joni_ is now known as alex_joni
[13:49:18] <les_w> morning/afternoon
[13:49:48] <les_w> I see josh compiled a nice list of TP papers
[13:49:50] <Jacky^> morning les_w
[13:49:58] <les_w> ha jacky
[13:50:15] <les_w> hi
[13:53:11] <Jacky^> id like to wich printer has been used to print on steel http://www.veroececcarelli.com/images/piastrelle-acciaio.jpg
[13:53:56] <Jacky^> know*
[13:55:34] <Jacky^> cold be laser printer ? on steel ? O_O
[13:55:48] <Jacky^> or chemical process maybe ..
[13:56:32] <skunkworks> screen printed? sublimated?
[13:56:55] <Jacky^> skunkworks: what is ? never eard
[13:57:26] <Jacky^> I tried to google, got no lucky ..
[13:57:59] <skunkworks> sublimation is ink transfer by heat. you would print the ink onto a sheet and then heat transfer it to the material - I think the metal would have to be coated though
[13:58:16] <skunkworks> hold on - I think we do it here
[13:58:28] <Jacky^> hot transfer ?
[13:59:05] <Jacky^> if that can print on steel, it can print pcb too
[13:59:26] <Jacky^> can print everywhere !
[14:00:13] <skunkworks> we do tiles like that simialar - talked to the guy - he thought that the steel would have to have a clear coat on it
[14:00:28] <skunkworks> it is called dye sublimation.
[14:01:07] <Jacky^> wow
[14:01:09] <skunkworks> we do custom tiles the same way - that was my first thought when I saw the photo
[14:01:35] <skunkworks> http://www.empirescreen.com/apersonalized.html
[14:02:19] <Jacky^> coool
[14:04:08] <skunkworks> I always wanted to try the laser printed transfer sheets for making pc boards
[14:04:53] <skunkworks> I was a dry transfer person my self. Some screen printed boards too when I wanted to make a bunch.
[14:06:10] <Jacky^> very interesting
[14:18:31] <jepler> skunkworks: about axis -- did you never consider that you might be able to rotate to a different view, or did you try some things but just not find the correct button to use to rotate?
[14:19:38] <alex_joni> jepler: what's wrong with that view?
[14:22:23] <skunkworks> You have to remeber I have only been using it for a week or so. ;) give me a break here. Plus I don't have a middle mouse button and I don't read direction.
[14:24:21] <jepler> alex_joni: Nothing; we hope the preset views are all useful. But rotating the view is also very useful.
[14:24:54] <cradek> skunkworks: he's not picking on you, he's trying to figure out how a new user learns to use the software
[14:24:58] <alex_joni> looks great to me
[14:25:18] <cradek> skunkworks: we think that if you didn't figure it out right away, it might mean the software is deficient in some way
[14:25:25] <jepler> not the user :-P
[14:25:34] <alex_joni> the developers :)
[14:25:37] <alex_joni> * alex_joni hides
[14:25:39] <jepler> all our users are perfectly .. ficient
[14:26:12] <jepler> if we thought it was important that you read the documentation before starting, we would have written 100 pages of it at a minimum
[14:26:35] <jepler> so we could rub your nose in the text just below the heading on page 21 that reads "In case of fire's"
[14:26:38] <alex_joni> so that they would have given up after 10..
[14:26:44] <skunkworks> I was not taking offence ;)
[14:26:49] <jepler> oh good
[14:27:34] <skunkworks> Had not really thought about it actually - when cradek said he tried to rotate it the little light bulb turned on.
[14:28:08] <skunkworks> Is there a key you can hold down for the non 3 button mouse persons?
[14:28:31] <cradek> skunkworks: no
[14:28:34] <jepler> skunkworks: no, there's not. But if your X server is properly configured, just hold down both buttons at the same time ("chord middle").
[14:28:55] <jepler> We assumed most of our users would have a wheel mouse, which can be clicked/held to rotate
[14:29:05] <alex_joni> jepler: how about this: another button for rotate (next to the P one), and then some ellipses appear
[14:29:08] <skunkworks> I will give that a try.
[14:29:08] <cradek> skunkworks: all X applications pretty much depend on having three buttons available, and as jepler says, two-buttoners have to emulate the third
[14:29:36] <skunkworks> again - I may be up to 3 weeks using linux ;)
[14:30:46] <cradek> skunkworks: if you need help turning on middle button chording, let one of us know
[14:30:56] <cradek> skunkworks: it was probably an option when you installed, but easy to miss
[14:31:06] <skunkworks> I brought the computer to work - booting it now.
[14:31:31] <jepler> cradek: what do you think about making modified LMB rotate?
[14:32:17] <cradek> jepler: fine with me; I suggest SHIFT so it has a chance of always working
[14:32:36] <skunkworks> when the k interface first came up on the bdi install I picked the microsoft emulation - that might have screwed me maybe?
[14:32:58] <cradek> skunkworks: not sure
[14:33:01] <chinamill> Does anyone have a .hal file with a parallel port input, declared and attached to function in emc2 I can learn from?
[14:34:19] <jepler> cradek: I'll stick that in
[14:40:20] <jepler> skunkworks: In the next version of axis you'll be able to rotate with the shift + left button.
[14:40:47] <jepler> skunkworks: if you want, you can make a small change in your copy of axis, and then "setup.py install" again. The change is shown in this file: http://axis.unpy.net/index.cgi-files/sandbox/rotate-with-shift-left-mousebutton.patch
[14:40:54] <alex_joni> jepler: do you know how rotation in autocad 3D works?
[14:41:02] <jepler> skunkworks: insert the two lines marked "+" in the file rs274/OpenGLTk.py near line 265
[14:41:05] <jepler> alex_joni: no
[14:41:15] <jepler> alex_joni: I don't own autocad and I've only seen cradek use a very old version (r12 for dos)
[14:41:17] <alex_joni> * alex_joni looks for a screenshot
[14:41:47] <jepler> skunkworks: Or download it and use the program 'patch': 'patch < rotate-*.patch' at the command prompt
[14:44:34] <skunkworks> Ok - if I click mouse buttons and then let up on the right buttin I get the rotate. Yes I am sure I would have figured that out ;)
[14:44:59] <skunkworks> damn thats nice
[14:46:02] <skunkworks> let me say that again if I click "Both" mouse buttons and then let up on the right buttin I get the rotate.
[14:47:16] <alex_joni> jepler: can you dcc?
[14:48:32] <jepler> alex_joni: it's been ages since I tried
[14:48:39] <cradek> works for me
[14:48:47] <alex_joni> I sent the request.. ok sending to chris
[14:49:16] <alex_joni> anything getting through?
[14:49:23] <alex_joni> if not I'll mail
[14:49:23] <cradek> just the request
[14:49:33] <cradek> you must not be able to send
[14:49:39] <alex_joni> ok.. mail then
[14:49:43] <jepler> 08:49:33 DCC no file offered by alex_joni
[14:50:19] <alex_joni> I closed them
[14:50:21] <alex_joni> and mailed chris
[14:50:37] <jepler> OK
[14:51:18] <alex_joni> btw, showed skunkworks's screenshot to some friends.. they were extatic about it ;)
[14:51:33] <jepler> which screenshot did skunkworks make?
[14:51:49] <cradek> jepler: ~/autocad?.jpg
[14:51:57] <alex_joni> http://www.electronicsam.com/images/KandT/AXIS.JPG
[14:51:59] <jepler> EPERM
[14:52:01] <cradek> shuttlecraft approaching for landing, sir
[14:52:11] <cradek> EAGAIN
[14:52:31] <alex_joni> there are some explanations for those pics..
[14:52:31] <jepler> what's the green smudge?
[14:52:54] <cradek> something about a pointer I think
[14:53:00] <alex_joni> ok.. the green circle is for rotation help
[14:53:06] <cradek> looks like a fish with earmuffs
[14:53:06] <alex_joni> see the 4 little green circles?
[14:53:19] <alex_joni> top/down, left/right?
[14:53:21] <cradek> yeah
[14:53:37] <alex_joni> when I enter the mouse in one of those, the mouse pointer changes
[14:53:48] <alex_joni> like in pic 2 and 3
[14:54:06] <alex_joni> pic 2 I visited one of the circles from the top or bottom
[14:54:08] <cradek> interesting
[14:54:31] <jepler> so after you put your mouse in the bottom circle you only rotate along one axis as shown by the mouse pointer?
[14:54:32] <alex_joni> when I drag the mouse (need to press inside one of the little circles), then it keeps one rotational axis fixed
[14:54:37] <alex_joni> exactly
[14:54:47] <cradek> I like our nav better then
[14:54:50] <alex_joni> if I drag inside the big circle it behaves just like axis
[14:55:04] <cradek> you can rotate in latitude, longitude simultaneously
[14:55:06] <cradek> oh
[14:55:07] <cradek> I see
[14:55:08] <alex_joni> and if I drag outside the big circle it rotates around the viewpoint
[14:55:18] <alex_joni> like in pic 4
[14:55:22] <jepler> rotating is two steps -- first, deploy the green circle, and then rotate in several different ways depending where you click?
[14:55:35] <alex_joni> let me say again:
[14:56:03] <alex_joni> 1. click on the bar on the top, just below Adobe PDF and to the right
[14:56:07] <alex_joni> there is a green world symbol
[14:56:24] <alex_joni> like a planet with a satelite around it.. do you see it?
[14:56:28] <jepler> yes
[14:56:42] <alex_joni> left to panning. ok , that enters 3D orbit
[14:56:49] <alex_joni> it's called in autocad at least
[14:57:03] <alex_joni> now.. inside the big green circle rotation works like in AXIS
[14:57:15] <jepler> do the left and right circles small circles do the same thing?
[14:57:20] <alex_joni> yes
[14:58:40] <alex_joni> want me to set up a VNC to try this?
[14:58:56] <jepler> no thanks, your description is pretty good
[14:59:12] <alex_joni> glad you understood..
[14:59:20] <skunkworks> Played with it some more - you guys did an awsome job. I could get used to the "cord middle" (although is mine not working 100% as I have to press both buttons and then let up on the right?)
[14:59:37] <alex_joni> just wanted to point out there are other ways for easy rotate..
[14:59:51] <alex_joni> I'm not saying I don't like how it works in AXIS now..
[15:00:10] <jepler> my reaction is that it looks really powerful, but more complicated
[15:00:25] <cradek> skunkworks: you should be able to press and hold both buttons
[15:00:56] <alex_joni> jeff: not something urgent, maybe to keep in the back of your head..
[15:00:56] <cradek> skunkworks: I think your mouse configuration is probably somehow wrong
[15:00:59] <skunkworks> hmmm - might be just how I initially set up the gui (microsoft windows emulation).
[15:01:19] <skunkworks> I will play with it. But yes I can rotate. nice.
[15:01:25] <cradek> skunkworks: probably, it's earlier, when you told it what kind of mouse you have (or it auto-detected wrongly)
[15:01:42] <jepler> skunkworks: try it with the patch I gave the URL for earlier. that'll let you use shift+left instead of the two buttons together.
[15:02:07] <alex_joni> jepler: also notice that there are a lot of views in autocad on the bottom of the screen
[15:02:24] <skunkworks> I will - it will be a bit though.
[15:02:51] <cradek> I don't understand what they mean
[15:02:58] <cradek> I think I understand the first six directions
[15:03:04] <cradek> but what are the additional 4 cubes?
[15:03:08] <jepler> cradek: the rest are 3/4 or perspective views, I think
[15:03:14] <alex_joni> basicly the same ones as in axis, but for P there is SW,SE,NW,NE
[15:03:24] <skunkworks> trying to think - I thing the mouse that I used to install the bdi had a center scroll. I will look around for a similar mouse and try it.
[15:03:29] <alex_joni> 4 perspectives..
[15:03:31] <cradek> ok
[15:03:53] <cradek> it's hard to express that in 20 pixels square
[15:04:02] <alex_joni> for all 4 Z is facing up
[15:04:11] <alex_joni> only XY rotate in 90 degree angles
[15:04:31] <jepler> cradek: axis' icons are 4 times that big and we still do a poor job
[15:04:55] <alex_joni> and there is an additional camera (which you can use to define the location of the viewport), but I hate that
[15:05:32] <alex_joni> sometimes words say a LOT more than pictures.. so tool-tips are a MUST
[15:09:17] <alex_joni> * alex_joni goes back to his stuff..
[15:09:33] <alex_joni> btw, I'm welding that part ;)
[15:09:34] <jepler> alex_joni: thanks for showing us that
[15:09:41] <alex_joni> jepler: anytime ;)
[15:09:54] <alex_joni> it's about 2-3m long
[15:10:02] <alex_joni> big turntable and robot :D
[15:10:11] <cradek> so it is a shuttlecraft
[15:10:24] <alex_joni> it's a frame for a construction machine ;)
[15:10:31] <alex_joni> excavator to be exact
[15:11:00] <alex_joni> in the first pic you can see 2 holes at it's tip
[15:13:36] <cradek> jeez
[15:13:46] <alex_joni> seen the pic I sent you?
[15:13:48] <cradek> yes
[15:13:55] <cradek> where does the part go?
[15:14:05] <alex_joni> if you look closely at the bottom of the frame
[15:14:10] <alex_joni> not frame.. arm
[15:14:29] <alex_joni> do you see the piston coming out of the body?
[15:14:31] <cradek> yes
[15:14:48] <alex_joni> next to the piston are the 2 holes I was talking about
[15:14:57] <cradek> gotcha
[15:14:58] <alex_joni> supporting the arm
[15:15:15] <alex_joni> so that part actually rotates with the cabin
[15:16:00] <alex_joni> pic 2 shows a nice view of the rotational axis, the funny hole
[15:41:11] <Jacky^> later
[15:41:20] <Jacky^> Jacky^ is now known as Jacky^afk
[16:32:17] <jepler> cradek: I checked in a new shift-lmb patch
[16:34:14] <cradek> jepler: thanks
[16:35:44] <skunkworks> thats not right - if you click on your name it goes to holmen - I guess it is isn't using the full accress
[16:35:49] <skunkworks> area code
[16:36:12] <skunkworks> sorry - wrong window - strike that
[16:40:17] <dave-e> ok, lots of people on this morning, anyone awake
[16:41:24] <skunkworks> I would like to say that I am not - but I am.
[16:41:32] <dave-e> even yet
[16:41:54] <dave-e> too early?? too late??
[16:42:47] <skunkworks> almost lunch time for me.
[16:43:25] <dave-e> ah... well I'm having morining coffee
[16:44:00] <dave-e> I do need an idea from someone....
[16:44:16] <dave-e> emc doesn't want to come up this morning
[16:44:25] <dave-e> generic will run
[16:44:36] <dave-e> xemc | tkemc will not come up
[16:45:04] <dave-e> it looks like primary shmem isn't happening... and emc.nml is there
[16:45:11] <les_w> hi dave
[16:45:18] <skunkworks> I am probabley not the person - 3 weeks so far on linux/emc2/axis ;)
[16:45:19] <dave-e> hi les
[16:45:24] <les_w> can't help but have a question
[16:45:28] <dave-e> ok
[16:45:50] <les_w> you lock up at 125 microsecond with a vital card?
[16:45:56] <les_w> servo rate
[16:46:02] <dave-e> haven't tried
[16:46:12] <dave-e> I do lock at 125 us with stg
[16:46:27] <dave-e> in fact that is what i'm trying to recover from
[16:46:37] <les_w> oh
[16:46:46] <les_w> can you run at 200?
[16:46:51] <dave-e> btw-Josh is running the 4 axis motion lite card
[16:47:02] <les_w> really
[16:47:26] <les_w> you know I had a long talk with fred yesterday.
[16:47:39] <dave-e> I don't know I jumped from 250 to 125 ....binary search ;-) ya know
[16:47:48] <dave-e> and
[16:47:48] <les_w> heh
[16:48:06] <SWPadnos> next try 187.5
[16:48:10] <SWPadnos> :)
[16:48:15] <les_w> The tc/tp planner doe not velocity adapt at all
[16:48:20] <les_w> it can't
[16:48:26] <les_w> that explains a lot
[16:48:47] <dave-e> any hooks there to do it
[16:49:03] <SWPadnos> the hook is that the TP is a separate source file
[16:49:08] <dave-e> hi swp
[16:49:11] <SWPadnos> which can be thrown out
[16:49:13] <SWPadnos> hi
[16:49:26] <les_w> so...any unrealizable motion command is going to make a mess
[16:49:36] <dave-e> indeed
[16:49:57] <les_w> that is why the queue gets drained
[16:50:12] <les_w> and why high servo update helps
[16:50:13] <dave-e> I was going to try long tp's this morning and see if the sub interp could do better
[16:50:41] <les_w> looks like file system got trashed somewhere in the lock up huh
[16:50:52] <dave-e> don't know
[16:50:59] <dave-e> emc.nml looks ok
[16:51:16] <les_w> must have if reboot didn't help
[16:51:32] <dave-e> maybe i should copy generic.nml across since it works
[16:51:53] <les_w> yeah
[16:52:12] <dave-e> so high servo rate helps...any other ideas
[16:52:22] <SWPadnos> what messages do you get?
[16:52:33] <les_w> I keep a "spare" emc in another directory
[16:52:35] <SWPadnos> it could also be something in the tcl directory is fubared
[16:52:56] <dave-e> something about not master blah blah...
[16:53:07] <dave-e> no communications established
[16:53:16] <SWPadnos> ok - that does sound more NML-ish than TCL-ish
[16:53:43] <dave-e> indeed it usually happens when one doesn't have a nml file
[16:54:07] <SWPadnos> this is emc1?
[16:54:12] <dave-e> yes
[16:54:53] <SWPadnos> ok - I'm not sure how the tcl programs get the nml file, but that's also something that could be wrong (incorrect cmd line options or environment)
[16:55:01] <dave-e> alex btw i checked the stg site just after you posted and could find no ref to new driver...am i just blind
[16:55:24] <alex_joni> dave-e: the driver is in emc2, not on the stg site
[16:55:31] <dave-e> ah.
[16:55:36] <dave-e> duh
[16:55:45] <alex_joni> I just posted the STG site as reference for people who don't know what I'm talking about ..
[16:55:51] <dave-e> ok
[16:55:52] <alex_joni> maybe I wasn't clear enough :/
[16:56:00] <les_w> dave what computer are you using?
[16:56:01] <alex_joni> sorry
[16:56:24] <dave-e> maybe they would put a pointer on their site to emc2 because of the driver
[16:56:45] <alex_joni> I'll mail them about it
[16:56:50] <alex_joni> good idea..
[16:57:16] <dave-e> I'm going to go back to the shop... and play some more. see ya later
[17:37:24] <Jymmm> It's surprising how many (including Celine Dion) can really fuckup a song like Ave Maria
[17:39:32] <Jacky^afk> hey Jymmm :)
[17:39:43] <Jymmm> Hi Jacky^afk
[17:39:51] <SWPadnos> Celine Dion tends to screw up just about any song
[17:39:52] <Jacky^afk> (nick Jacky^
[17:39:57] <Jacky^afk> Jacky^afk is now known as Jacky^
[17:40:10] <Jymmm> SWPadnos no doubt.
[17:40:21] <Jacky^> :-))
[17:40:34] <SWPadnos> I first started to dislike her when I heard the cover of "All By Myself"
[17:40:46] <Jymmm> lol
[17:40:57] <Jacky^> :P
[18:03:10] <skunkworks> chinamill?
[18:30:02] <les_w> blah. taking a break. Didn't have a 5/8 reamer, and am hving to use the boring head for a hole. Time waster!
[18:31:02] <les_w> lerman: you on?
[18:32:28] <les_w> Was thinking about non-RT planners.
[18:32:34] <Jymmm> * Jymmm hands les_w a blowtorch
[18:33:44] <les_w> funny thomson bearings have a .0005 tol for the mount hole, but have 10 times the slop on the shaft
[18:35:56] <Jymmm> I meant to ask you... how often tdo you lube the rails on your router?
[18:36:21] <les_w> about twice a year
[18:36:43] <les_w> z more often because it's exposed
[18:36:47] <Jymmm> even with dust getting on them ?
[18:37:03] <les_w> woops customer
[18:37:15] <Jymmm> k
[19:35:38] <Jacky^> hello :)
[19:35:57] <rayh> Hello Jacky^
[19:36:22] <Jacky^> hi rayh !
[19:37:05] <Jacky^> rayh: how are the things going there ?
[19:38:14] <rayh> Good. A bit cool but good.
[19:38:25] <Jacky^> ;-)
[19:39:19] <rayh> How about you. Are things good there?
[19:39:35] <Jacky^> uhm.. not at all
[19:39:51] <Jacky^> but what we can do ? nothing
[19:39:56] <Jacky^> just wait
[19:39:58] <Jacky^> :)
[19:40:34] <Jacky^> 2005 was a very bad year here
[19:40:47] <rayh> Why is that?
[19:40:50] <Jacky^> hope new year will change something
[19:41:39] <Jacky^> oh, just why peoples think bad
[19:41:53] <Jacky^> or dont think at all
[19:42:07] <Jacky^> waht is really important and what not
[19:42:14] <rayh> I think I know what you mean.
[19:42:25] <Jacky^> im sure you know
[19:42:28] <Jacky^> :)
[19:42:55] <Jymmm> all the peoples think Jacky^ in a dress is bad =)
[19:43:29] <rayh> okay...
[19:43:30] <Jacky^> Jymmm: my latest hope is the net
[19:43:43] <Jacky^> i believe in the net
[19:43:52] <Jacky^> is the only thing
[19:44:07] <Jymmm> Jacky^: fish net stockings?
[19:44:24] <Jacky^> heheh
[19:44:25] <alex_joni> undernet
[19:44:33] <Jymmm> underlord
[19:44:37] <Jacky^> no, I meant ..
[19:44:38] <alex_joni> alternet
[19:44:45] <Jymmm> dragnet
[19:44:52] <Jacky^> with the net we can do democracy directly
[19:44:57] <Jacky^> aone to one
[19:45:18] <Jacky^> I believe in that
[19:45:30] <Jymmm> Jacky^ internet has opened up a lot of possibilities, though ppl will always suck in one form or another.
[19:46:02] <Jymmm> Though I doubt I could function w/o it.
[19:47:10] <Jacky^> well, I want every people born, he born with an Internet Card Identity for free
[19:47:32] <Jacky^> knowledge for free fro all
[19:47:36] <Jymmm> I'd rather stay anonymous
[19:47:37] <Jacky^> for*
[19:48:09] <Jacky^> Jymmm: wrong ..
[19:50:46] <Jacky^> it can really change the world
[19:50:58] <Jacky^> and it will be
[19:54:27] <Jacky^> we cant go on doing wars
[19:55:01] <Jacky^> the only way is to export our democracy all over the world
[19:55:23] <Jacky^> whan peoples understand this, understand all
[19:55:56] <Jacky^> and we'll be happy
[19:55:59] <CIA-12> 03cradek * 10emc2/src/emc/ (13 files in 5 dirs):
[19:55:59] <CIA-12> This is most of the fix for bug 1377336: AXIS velocity constraints are
[19:56:00] <CIA-12> ignored by Feed Override.
[19:56:00] <CIA-12> I think arcs are not yet right.
[19:56:56] <Jacky^> but we cant export our democracy with the guns
[19:57:11] <Jacky^> its wrong !
[19:57:24] <cradek> and it fails
[19:57:32] <dmess> you CaNT EXPORT Democrocy
[19:57:40] <Jacky^> we can
[19:57:47] <dmess> we??
[19:58:00] <Jacky^> democracy mean knowledge
[19:58:06] <Jacky^> yes
[19:58:40] <dmess> sorry ... 1/2 way in conversation
[19:59:01] <Jacky^> we could expoert emc in china
[19:59:23] <Jacky^> but maybe a guy googling emc in china do not find nothing !
[19:59:26] <Jacky^> about emc
[19:59:37] <dmess> oh i dont want my fingerprint on that train wreck
[20:00:22] <Jacky^> you know why ?
[20:00:47] <Jacky^> because china government payed google to censure results on search
[20:00:56] <Jacky^> this is not democracy !
[20:01:43] <Jacky^> that what we can do
[20:02:09] <SWPadnos> democracy is a method of choosing a government, it has nothing to do with censure
[20:02:18] <dmess> certain things in the wrong hands ... will mean IMMEDIATE retaliatory military action
[20:03:01] <Jacky^> ok, but the story, should be teach something
[20:03:08] <Jacky^> we know the story
[20:03:19] <Jacky^> i know the story of italy
[20:03:29] <Jacky^> I cant forget that
[20:04:05] <Jacky^> so, I think we have the power in our hands
[20:04:16] <Jacky^> but were not using it at all
[20:05:48] <dmess> you got it boy.... the cars moving ... but nobody is driving it
[20:06:04] <Jacky^> I cant accept that
[20:06:21] <Jacky^> at least .. if Im alive ..
[20:06:49] <dmess> i dont... i try to drive the engine.... HARD
[20:06:50] <Jacky^> i'm a rebel then
[20:08:17] <dmess> ive been TOLD to lie down and die too many times ... now i just dont listen anymore..
[20:08:41] <Jacky^> that what all peoples think
[20:09:02] <dmess> not like me
[20:10:09] <Jacky^> Id like to know why peoples from a lot of countries are not here around
[20:11:02] <Jacky^> why some government close peoples blogs and dont permit peoples to talk
[20:11:10] <Jacky^> a lot
[20:11:15] <dmess> they have their own old military sotware to play with
[20:11:37] <Jacky^> well, after wi won on irak
[20:11:43] <Jacky^> what will change ?
[20:11:50] <Jacky^> in the world ?
[20:11:53] <k4ts> hello
[20:11:57] <Jacky^> hi k4ts
[20:11:59] <Jacky^> :)
[20:13:22] <Jacky^> k4ts: talking about free and democracy
[20:13:29] <Jacky^> freedom
[20:14:29] <Jacky^> k4ts: are U ?
[20:15:19] <k4ts> I'm
[20:15:20] <k4ts> ah ah
[20:15:29] <k4ts> qui!
[20:15:32] <Jacky^> oh :)
[20:15:45] <Jacky^> good dinner ?
[20:16:05] <k4ts> yes
[20:16:12] <k4ts> bleach
[20:16:21] <Jacky^> what ?
[20:16:33] <k4ts> cotoletta e fagiolini
[20:16:38] <Jacky^> bauahahahaha
[20:16:43] <Jacky^> nicee
[20:16:59] <k4ts> avrei preferito spaghetti a vongole
[20:17:08] <Jacky^> :-)
[20:17:11] <k4ts> :-)
[20:17:21] <Jacky^> good
[20:18:55] <Jacky^> hi martin :)
[20:19:32] <Imperator_> Hi Jacky^
[20:19:53] <Imperator_> SOMETHING NEW ?
[20:20:32] <Jacky^> oh.. yes, I think the guiys here have a lot of news about TP
[20:20:44] <Jacky^> have you been in the latest discussions ?
[20:20:54] <Jacky^> guys*
[20:21:42] <Imperator_> nope sorry
[20:22:15] <Imperator_> but im very interestet in that stuff, because i would like to build a high speed machine
[20:22:33] <Jacky^> yeah, nice :)
[20:22:52] <Jacky^> I guess theres some progress
[20:23:35] <Jacky^> Im not following tecnical topics at all in these days
[20:24:07] <Imperator_> :-)
[20:24:09] <Jacky^> taking a pause
[20:24:12] <Jacky^> ;)
[20:24:33] <Imperator_> i worked the last years with s HSC machine from Digma
[20:24:43] <Imperator_> that was a nice toy
[20:25:46] <CIA-12> 03cradek * 10emc2/src/emc/task/emccanon.cc:
[20:25:46] <CIA-12> same fix for arcs (bug 1377336: AXIS_* velocity constraints are ignored
[20:25:46] <CIA-12> by Feed Override)
[20:28:41] <jepler> yay!
[20:28:55] <cradek> whee
[20:29:36] <SWPadnos> does that slow an entire arc, or will it move at different rates depending on the actuall motion being executed?
[20:29:40] <SWPadnos> oh - and wheeee!
[20:29:49] <cradek> it slows the entire arc
[20:29:55] <SWPadnos> ok
[20:36:07] <alex_joni> wheee
[20:38:40] <Jacky^> * Jacky^ k4ts and Jack in paltalk #Italiani e Brasiliani abbracciati nel mondo!
[20:38:43] <Jacky^> :P
[20:45:13] <Jacky^> hahahah
[20:48:03] <Jymmm> anyone know where to get the plastic caps that cover the bolt mounting holes for linear rails?
[20:48:28] <Jacky^> ugh
[20:48:47] <Jacky^> I think is expensive
[20:49:11] <Jacky^> and not useful at all
[20:49:45] <Jymmm> Jacky^: They keep debri from entering the slides.
[20:49:58] <Jacky^> Jymmm: understood
[20:52:12] <Jacky^> :))
[20:53:04] <Jacky^> where is k4ts ?
[20:56:30] <k4ts> hello
[20:56:39] <Jacky^> he !
[20:56:43] <Jacky^> :)
[20:57:05] <k4ts> no stonato � che � la seconda volta che la canto
[20:57:13] <k4ts> e non la conosco bene
[20:57:15] <Jacky^> heheh ok
[21:10:41] <cradek> jepler: I notice nobody's testing it...
[21:11:12] <alex_joni_> * alex_joni_ did
[21:11:38] <cradek> did you do arcs?
[21:11:45] <cradek> (I didn't either)
[21:11:53] <alex_joni_> no arcs ;)
[21:12:22] <alex_joni_> cvs up, for the latest, had some conflicts on merge
[21:12:35] <cradek> I notice that arc velocity calculations are wrong... in many cases it could take them faster
[21:12:44] <cradek> I mean, even before I started
[21:14:01] <cradek> alex_joni_: just nuke those files and up again
[21:14:43] <alex_joni_> I did cvs -dPC
[21:26:28] <dmwaters> {global notice} Hi all, that split was my fault. had to update the server's firewall, and it apparrently reset the connections on the server. I apologize for this, and thank you for using freenode!
[21:37:42] <alex_joni> night all
[21:43:20] <cradek_> cradek_ is now known as cradek
[21:43:53] <cradek> I think freenode isn't doing so well today
[21:46:14] <alex_joni> oh my.. massive reconnects ;)
[21:47:21] <Jymmm> alex_joni: firewall was rebooted.
[21:47:48] <alex_joni> was there a wallop?
[21:47:55] <Jymmm> yep
[21:50:14] <alex_joni> didn't get through to here
[22:15:03] <anonimas1> alex_joni: drop the mail ^_^
[22:15:22] <alex_joni> I did
[22:15:25] <alex_joni> a while ago
[22:29:46] <CIA-12> 03alex_joni * 10emc2/src/emc/task/emccanon.cc: fixed a minor bug that caused the proper velocity to not get printed for arcs. arcs do work as promised
[22:36:50] <anonimas1> anonimas1 is now known as anonimasu
[22:46:33] <anonimasu> logger_aj: bookmark
[22:46:34] <anonimasu> See http://solaris.cs.utt.ro/irc/irc.freenode.net:6667/emc/2005-12-09#T22-46-33
[22:54:16] <chinamill> Hmmm, has anyone got a rapid movement problem with emc2? It seemes when doing G0 the axis goes a little bit too far, and then backs up to the ordered position. (CVS update one week ago)
[22:54:37] <alex_joni> try a more recent one
[22:55:33] <fenn_> woohoo we may have an open source CAM package coming soon
[22:55:41] <chinamill> Can problems like that come of a weird ini?
[22:56:22] <chinamill> fenn: wich one?
[22:56:39] <SWPadnos_> BRL_CAD?
[22:57:02] <chinamill> I have seen some test with blender
[22:57:16] <fenn_> chinamill: i wrote to the guy behind freesteel.co.uk and he says he would like help on making a user interface to some of his algorithms
[22:57:28] <chinamill> (only surface modeling)
[22:57:29] <fenn_> brlcad and blender are cad programs
[22:57:30] <SWPadnos_> cool
[22:57:35] <fenn_> i'm talking about toolpath planning
[22:57:54] <SWPadnos_> check with Rab Gordon of "CNC Tollkit" fame
[22:58:04] <fenn_> that's all integrated with 3d max
[22:58:18] <SWPadnos_> he's looking for a package to base it on, since GMax is no longer free
[22:58:20] <chinamill> Rabs package is not open source...
[22:58:24] <SWPadnos_> no, it's not
[22:58:38] <SWPadnos_> though his does 5 axis toolpaths
[22:59:05] <chinamill> The tests with blender generates g-code but it is not yet ad fancy as "CNC toolkit"
[22:59:24] <fenn_> wow really
[22:59:43] <fenn_> last time i checked out blender they were still trying to get numeric input
[22:59:52] <fenn_> instead of dragging the mouse around
[23:00:13] <SWPadnos_> it's got an interface only a mother could love, but it is quite powerful (from what I hear - I have nearly no experience with it)
[23:00:26] <fenn_> yeah blender is a pain in the ass to use
[23:00:46] <fenn_> worse than brlcad
[23:00:49] <chinamill> Yes, blender is more of a tool to model stuff that looks like the real thing... not yet for mechanical CAD
[23:01:10] <chinamill> what is bricad?
[23:01:33] <SWPadnos_> brl-cad is the open source CAD package originally made by the army
[23:01:34] <fenn_> a solid modeler the army uses for tank simulations
[23:01:56] <SWPadnos_> it does a lot of stuff, but has an interface only a grandmother could love ;)
[23:02:23] <chinamill> :)
[23:08:31] <SWPadnos_> in fact, I couldn't get it to give me an interface on this machine
[23:08:32] <SWPadnos_> may be a 64-bit or driver problem
[23:08:32] <fenn_> i would be happy with: nurbs surfaces, csg, kinematic constraints, and relational constraints
[23:08:32] <SWPadnos_> you could make your own kitchen sink :)
[23:08:32] <fenn_> yep
[23:08:32] <fenn_> no fluid dynamics though
[23:08:32] <fenn_> actually, an FEA algorithm built into the modeler would be really nice
[23:08:32] <SWPadnos_> there's a FEA package that's open source, but I'm not sure about fluid diynamics
[23:08:32] <SWPadnos_> I thikn brl-cad does FEA, actually
[23:08:32] <fenn_> i havent been able to penetrate the brl documentation
[23:08:32] <chinamill> Anyone knows bricads homepage? (cant find it)
[23:08:32] <fenn_> brlcad.org
[23:08:32] <SWPadnos_> www.brlcad.org ;)
[23:08:32] <SWPadnos_> that's an "L", not an "I"
[23:08:32] <fenn_> heh
[23:08:32] <fenn_> ray pronounces it BRILL-CAD
[23:08:32] <SWPadnos_> yep - that's how I would say it
[23:08:32] <fenn_> do you say fur filter or F-I-R filter
[23:08:32] <SWPadnos_> F-I-R
[23:08:32] <SWPadnos_> eff eye arrrrr
[23:08:32] <fenn_> matey
[23:08:32] <SWPadnos_> that would be eff aye arrrrr
[23:15:16] <chinamill> Anyone got a .HAL with a parallel port input attached to a function in ecm2, that I can learn from?
[23:15:54] <alex_joni> attached to a function?
[23:16:34] <chinamill> Something like a E-stop controlled by a input on the parport
[23:16:42] <SWPadnos_> look at core_stepper.hal - that has a lot of connections
[23:17:03] <chinamill> not parport inputs
[23:17:28] <alex_joni> standard_pinout.hal is for parport
[23:17:58] <alex_joni> and look at http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl/emcinfo.pl?Sample_HAL_And_ClassicLadder
[23:18:37] <SWPadnos_> actually - standard_pinout doesn't connect anything to the parport inputs either ;)
[23:18:58] <chinamill> I got spindle_on to work with a second parport, but I couldent figure out how to use parport input and to attack to the E-stop (function)
[23:19:39] <alex_joni> for ESTOP with external estop you need to use classicladder
[23:19:54] <chinamill> need or must?
[23:20:10] <fenn_> if you want it to stay stopped after you press the button
[23:20:25] <chinamill> good idea...
[23:20:29] <SWPadnos_> can't you just use an AND block?
[23:20:36] <anonimasu> logger_aj: bookmark
[23:20:36] <anonimasu> See http://solaris.cs.utt.ro/irc/irc.freenode.net:6667/emc/2005-12-09#T23-20-36
[23:21:09] <fenn_> swp how do you get it out of estop then?
[23:21:45] <SWPadnos_> good question - I don't know exactly how the GUI / motion estop interlock works
[23:22:22] <SWPadnos_> but, as I understand it, there's an output from motion (or io) that says "the GUI is not in stop mode"
[23:22:51] <SWPadnos_> I think that's a direct connection to the GUI estop pin
[23:23:09] <alex_joni> rewind guys
[23:23:12] <alex_joni> there are 2 things
[23:23:18] <alex_joni> 1. estop out from emc to hardware
[23:23:24] <alex_joni> 2. estop in from hardware to emc
[23:23:33] <alex_joni> in a simple system you simply connect those two together
[23:23:46] <alex_joni> like in core_stepper.hal
[23:23:47] <alex_joni> iirc
[23:23:57] <SWPadnos_> yes - that's what it dos
[23:23:58] <SWPadnos_> does
[23:24:09] <alex_joni> if you have an external estop chain, you need a something
[23:24:14] <alex_joni> to keep track of that
[23:24:20] <alex_joni> and the easiest is classicladder
[23:24:24] <alex_joni> because it'll do latching
[23:24:32] <SWPadnos_> I'm not sure that's the easiest
[23:24:33] <alex_joni> check the wiki I posted, case 4 iirc
[23:24:47] <anonimasu> alex_joni: I am reading back on the logs now..
[23:24:50] <alex_joni> maybe not the easiest..
[23:25:17] <SWPadnos_> and a heck of a lot more confusing when you look at the list of HAL pins ;)
[23:25:44] <chinamill> In general is emc2 alot more hardware demanding than emc1?
[23:25:52] <alex_joni> no
[23:25:59] <SWPadnos_> you can connect the parport input and the GUI output to an AND2 block, and connect the output to the motion.enable input, I think
[23:26:02] <alex_joni> confusing? why?
[23:26:12] <SWPadnos_> because there are 100 more pins to look at
[23:26:30] <alex_joni> yes.. if you don't mind that emc will go back on when external estop has cleared
[23:26:46] <alex_joni> not fully back on, but back to estop_reset
[23:27:12] <SWPadnos_> ok - so the GUI doesn't go into estop mode when it sees the estop happen?
[23:27:31] <SWPadnos_> (or, the gui doesn't stop telling emc that it's OK to run)
[23:27:57] <alex_joni> the gui isn't telling emc anything ;)
[23:28:05] <alex_joni> it's telling iocontrol which tells the hardware
[23:28:10] <alex_joni> which tells it back to iocontrol
[23:28:16] <alex_joni> which tells it to the gui
[23:28:20] <alex_joni> which sets the task state
[23:28:24] <alex_joni> clear now?
[23:28:28] <SWPadnos_> heh
[23:28:40] <chinamill> (less blury ;)
[23:29:15] <SWPadnos_> ok - there are definitely issues with that, and ladder takes care of them
[23:29:51] <SWPadnos_> ok - user-enable-out is momentary?
[23:29:52] <alex_joni> ladder can take care of all the versions you'd want to run..
[23:30:00] <alex_joni> it's the state of the GUI button
[23:30:06] <alex_joni> not momentary
[23:30:17] <alex_joni> hang on.. I might confuse the two ;)
[23:30:23] <SWPadnos_> heh
[23:30:32] <SWPadnos_> type 4 is a latching AND
[23:30:33] <alex_joni> and I wrote it..
[23:30:34] <alex_joni> :D
[23:31:51] <SWPadnos_> ok - I'd love to chat, but dinner with the wife sounds good too ;)
[23:32:01] <SWPadnos_> SWPadnos_ is now known as SWP_Away
[23:32:11] <alex_joni> ok.. enable out is the state of GUI button
[23:32:16] <alex_joni> and request-enable is momentary
[23:32:45] <SWPadnos> ok
[23:33:20] <SWPadnos> and request-enable happens when the GUI sends a ESTOP_OFF message (or whatever convoluted name it has)?
[23:33:36] <alex_joni> yes
[23:33:37] <chinamill> Nighti-nighti for me... bye
[23:33:40] <alex_joni> bye
[23:37:07] <anonimasu> 'yawns*
[23:37:17] <alex_joni> got the pics?
[23:50:08] <anonimasu> no
[23:51:42] <alex_joni> darn.. going to bed
[23:51:48] <alex_joni> I'll try tomorrow..ok?
[23:52:07] <anonimasu> 2sure
[23:52:08] <anonimasu> night
[23:52:16] <anonimasu> I'll be heading to bed in a bit also
[23:52:17] <anonimasu> :)
[23:52:20] <anonimasu> night
[23:52:23] <fenn_> nite nite
[23:53:09] <fenn_> pete what do you use to look at an xml file rendered with some particular stylesheet?
[23:56:44] <petev> fenn: I'm not an expert, but browsers seem to handle that
[23:57:07] <petev> I know explorer will and I bet others do too
[23:57:51] <fenn_> um, i guess a dtd is not a stylesheet. damn.
[23:58:07] <petev> no, DTD specifies what's lega;, not how to display it
[23:58:12] <petev> legal
[23:58:26] <petev> there are a few open source packages that handle DTDs
[23:58:42] <petev> but they are pretty heavy and are intended for full web development