#emc-devel | Logs for 2008-04-04

[02:04:34] <jepler> http://emergent.unpy.net/index.cgi-files/sandbox/demo_comp3.png
[02:06:01] <cradek> beautiful!
[02:06:17] <jepler> thanks
[02:06:47] <cradek> the un-delooped paths are hard to love...
[02:08:09] <cradek> I basically understand how to do islands - you just have to find loops (circuits that go the wrong way) consisting of more than one path
[02:08:23] <cradek> is that an easy or hard addition to your algorithm?
[02:08:43] <jepler> I'm not sure
[02:09:38] <jepler> the code is at http://git.unpy.net/view/comp.git
[02:10:03] <cradek> oh no, not another thing named comp!
[02:10:12] <jepler> yeah yeah
[02:10:30] <jepler> I'll rename it 'offs'
[02:10:38] <cradek> haha
[02:11:46] <cradek> I wonder how much of the same code we've written. I haven't looked at any of yours.
[02:12:07] <cradek> maybe we should continue to work separately; we may find and solve different problems
[02:13:15] <jepler> I hope there's no ill will about the idea that we might be duplicating effort
[02:13:26] <cradek> certainly not
[02:13:35] <cradek> I think we probably have different goals
[02:14:14] <jepler> officially renamed: http://git.unpy.net/view/offs.git
[02:14:29] <cradek> cool.
[02:15:26] <jepler> I don't know if we can have different goals if I'm unable to articulate my goal
[02:16:08] <jepler> (or maybe in that case we can't have the same goals :P)
[02:16:17] <cradek> a solution for emc's tool comp will have to be pretty tightly coupled to the interpreter. I have goals for your project (not strictly related to emc) even if you don't.
[02:16:52] <jepler> hah
[02:17:06] <cradek> I think you're going to end up with a good pocketing algorithm. the FOSS world doesn't have that yet.
[02:19:40] <jepler> the idea of what I have to do to handle islands and paths breaking apart is floating just out of reach. I know it involves examining the intersections (including those between different paths), and knowing when to discard a portion of a loop and when to make it into a new disconnected part..
[02:19:54] <cradek> yes
[02:20:07] <cradek> what part is out of reach?
[02:20:13] <jepler> umm, the details?
[02:20:19] <cradek> which ones?
[02:20:59] <cradek> (we probably need to doodle)
[02:21:22] <jepler> I don't feel like considering it too much right now
[02:21:34] <cradek> let's doodle tomorrow
[02:21:46] <jmkasunich> islands are chunks of metal that you want to leave, right?
[02:21:50] <jepler> jmkasunich: yes
[02:22:14] <jmkasunich> what about "bays" - chunks of metal that you want to remove, but that you can't reach from the "ocean"?
[02:22:14] <cradek> a path could cross itself many times - I'd have to draw some pictures to see if I'm right in thinking that the winding says whether to keep or discard
[02:22:25] <jepler> cradek: you should finish what you're doing with the interpreter first. I won't look at this again before monday night (I'm going out of town this weekend), so next week is a better time to doodle.
[02:22:35] <cradek> heh "finish"
[02:22:44] <jepler> I am pretty sure you're right about the winding
[02:22:49] <cradek> as if I know my goals there either
[02:23:32] <cradek> jmkasunich: there is no ocean, only one or more pockets...?
[02:23:39] <cradek> jmkasunich: I probably don't get your question
[02:23:54] <jmkasunich> the ocean is the pocket that you want to mill
[02:24:13] <jepler> jmkasunich: sometimes these algorithms postulate a vast coastline well outside the reach of the fastest ships...
[02:24:34] <jmkasunich> the bay is a section of that pocket connected to the main pocket by a passage that the tool won't fit thru
[02:24:48] <cradek> ok, that's just the path breaking apart
[02:25:04] <cradek> at some offset, that will happen for all but the simplest pockets
[02:25:08] <jmkasunich> are you discarding the part inside the bay? or doing two paths, one for each
[02:25:44] <cradek> at a certain offset your path breaks into two; the middle part flips winding and so is discarded
[02:26:25] <jmkasunich> would a more nitpicky description be that it breaks into three, you discard the middle, and join the other two?
[02:26:47] <cradek> it does break into three because there are two crossings.
[02:26:54] <cradek> I don't know what you mean by join though
[02:27:13] <jmkasunich> look at the 2nd pic down, on the left of jepler's image
[02:27:31] <jmkasunich> the U shaped part of the profile results in a loop
[02:27:31] <cradek> those aren't closed paths so it's a little tricky to talk about winding
[02:27:37] <cradek> yes
[02:28:06] <jmkasunich> the path is broken into three parts - the start, the end, and the middle - the middle is discarded, and the start is joined to the end where they cross
[02:28:30] <cradek> here is a much better pocket to consider http://timeguy.com/cradek-files/emc/winding.png
[02:29:12] <cradek> assuming the S is ccw (I'm sure it is) if you draw little arrows along the paths you will see the loops you keep are ccw, the ones you discard are cw
[02:29:21] <jmkasunich> I think I already see the why I'm seeing three and you see two - I have a start and an end, you have one continuous loop
[02:29:42] <jmkasunich> ok, got it
[02:30:06] <jmkasunich> the small loops at the ends of the S are what I'm calling bays (the big loop in the middle of the S is the ocean)
[02:30:11] <jepler> cradek: here's a case that would make my method throw fits if it was used as the general cutter comp algorithm in emc
[02:30:21] <cradek> for islands, the rule is the same (you make the island cw)
[02:30:39] <jepler> cradek: your "cut a slot in multiple passes" gcode would, since each pass has a bunch of coincident lines at varying "Z"
[02:31:08] <jepler> "ooh, I'd better throw something out -- but what?"
[02:31:12] <jmkasunich> heh
[02:31:20] <cradek> sorry, you lost me
[02:31:51] <cradek> oh you mean considering the whole path (which is stacked on top of itself) at once?
[02:31:55] <jepler> cradek: yes
[02:31:58] <cradek> yeah, ouch
[02:32:16] <jmkasunich> in general, I think the algorithm has problems when Z isn't constant
[02:32:25] <jepler> indeed
[02:32:26] <jepler> bbl
[02:32:28] <cradek> that is problematic all around
[02:32:32] <jmkasunich> but the cases where Z is constant are VERY usefull
[02:32:33] <jepler> you may continue to discuss my greatness in my absence
[02:32:36] <cradek> (I haven't handled it yet)
[02:33:01] <cradek> move into a (concave) corner, move Z up and down a thousand times, move to the next corner ...
[02:33:41] <jmkasunich> when you guys talk about this being pocketing code, you mean run the same profile with a number of diameters to generate the path to clear the entire pocket, right? (at the CAM stage)
[02:33:50] <cradek> yes
[02:33:54] <jmkasunich> cool
[02:33:57] <cradek> that's what I think jepler's code will be good for.
[02:34:14] <jmkasunich> start with actual tool diameter, then increase by 40% of dia (or whatever), until the entire path is being discarded
[02:34:15] <cradek> ccomp in emc won't be so good at that (and that's ok)
[02:34:22] <cradek> exactly
[02:34:52] <jmkasunich> I should stop chatting and start programming - I have a fun little part to machine
[02:35:06] <jmkasunich> intake manifold for a Tecumsah - a friend had 18 of them cast
[02:35:31] <cradek> do they warp or crack or something?
[02:35:38] <jmkasunich> I think fixturing it is the most complex setup I've done
[02:35:50] <jmkasunich> I dunno why he's doing it
[02:35:57] <cradek> I bet castings are always like that :-/
[02:36:08] <cradek> I've avoided that so far
[02:36:11] <cradek> you should take a photo
[02:36:27] <cradek> tecumseh is lawnmower/gocart type engines right?
[02:36:31] <jmkasunich> I have 5 clamps, 7 step blocks, 4 studs and t-nuts, 5 nuts, a vise, a parallel, and a little stack of cut/filed plates for indexing
[02:36:45] <jmkasunich> yes - I think in this case its mini-bikes
[02:37:02] <cradek> heh, is that all to flycut it flat?
[02:37:39] <jmkasunich> yes
[02:37:52] <jmkasunich> although I won't be flycutting - using a pattern with a 5/16" cutter
[02:38:04] <jmkasunich> my mill likes high speed light cuts better than slow stuff
[02:38:46] <jmkasunich> gonna flatten the mounting face and drill the mounting holes, plus clean up the inside of the port a bit
[02:38:56] <cradek> fun
[02:39:06] <cradek> do take a photo, I want to see your crazy setup
[02:39:29] <jmkasunich> just did, stand by
[02:40:35] <cradek> BOSS has a canned cycle to face with an endmill like that...
[02:40:39] <cradek> G183 iirc
[02:41:23] <cradek> G183 X Y Z X Y Z Z F F or somesuch nonsense
[02:41:26] <jmkasunich> http://jmkasunich.com/pics/IMGP1995.JPG
[02:41:52] <cradek> wow!
[02:42:15] <jmkasunich> some of the clamps are for convenience
[02:42:27] <jmkasunich> so that step blocks and such don't move when the actual part holding clamp is loosened
[02:42:44] <jmkasunich> if I wasn't looking at doing 18 pieces I would use something simpler
[02:42:56] <jmkasunich> but this locates to casting features and seems pretty solid
[02:43:11] <cradek> 18 is sure enough to justify a good setup
[02:43:27] <jmkasunich> the other end is gonna need machined too
[02:43:46] <jmkasunich> but for that end, we're thinking of screwing the first end to a plate, and setting the plate at the appropriate angle
[02:43:49] <cradek> this is intake for just one cylinder?
[02:44:01] <jmkasunich> yeah, its a one-lung engine
[02:44:17] <cradek> ah yep, then you get the angles right
[02:44:27] <cradek> (do you know the spec?)
[02:44:33] <jmkasunich> 20 degrees
[02:44:37] <jmkasunich> google rules: http://www.pbase.com/manilaman04/image/48318748/original.jpg
[02:44:58] <jmkasunich> you can see the manifold at the top of the engine, and the carb right below the fuel tank
[02:45:01] <cradek> not high precision work :-)
[02:45:15] <jmkasunich> yeah, 20 +/- 2 would probably be fine
[02:45:17] <cradek> the carb could move an inch in any direction!
[02:45:27] <cradek> maybe not quite, but close :-)
[02:45:32] <jmkasunich> heh
[02:45:55] <cradek> those must fail somehow if he's making these. interesting.
[02:46:17] <jmkasunich> I think he told me why, but that was a month or so ago
[02:46:22] <cradek> is this 2 stroke?
[02:46:36] <jmkasunich> I was in the middle of e-week madness when he got his castings
[02:47:06] <jmkasunich> not sure - it doesn't seem like it (at least it isn't using crankcase induction)
[02:48:58] <jmkasunich> hmm, that photo doesn't match up with the casting I have
[02:49:00] <cradek> so there are no holes in the ears of his castings at all?
[02:49:05] <jmkasunich> there must be multiple versions
[02:49:09] <jmkasunich> no
[02:49:26] <cradek> I don't understand how the other end mounts to the carb
[02:49:31] <jmkasunich> they're only 1/4-20 holes, way to small to core
[02:49:56] <jmkasunich> I think the end that I'm about to machine is the carb end
[02:50:09] <jmkasunich> screws go thru the carb ears into threaded holes in the manifold hears
[02:50:12] <jmkasunich> ears
[02:50:26] <jmkasunich> the other end has thru holes to mount to the block
[02:50:33] <cradek> oh the manifold is tapped, not the carb. interesting
[02:51:14] <jmkasunich> if i had a spindle VFD (and an encoder on the mill spindle) I could try rigid tapping ;-)
[02:51:29] <cradek> I'm sure anxious to do that on mine too!
[02:51:36] <cradek> it will be really nice.
[02:51:42] <jmkasunich> on the bport?
[02:51:45] <cradek> yes
[02:51:53] <jmkasunich> you gotta get busy with your conversion then
[02:52:07] <cradek> maybe this winter
[02:52:28] <cradek> I don't want to break it until the bus is about done
[02:52:33] <jmkasunich> yeah
[02:52:40] <cradek> too many things to do anyway
[02:52:43] <jmkasunich> I gotta get moving on this program - almost 11pm here
[02:52:48] <cradek> seeya then
[02:53:02] <jmkasunich> I want to do the one tonight, he's talking about bringing the others and coming over tomorrow
[02:53:23] <jmkasunich> (this is the guy that's going to NAMES with me in a couple weeks)
[02:53:34] <cradek> if it's already clamped, it will be a quick job now
[02:53:47] <jmkasunich> heh, just gotta write the program, and touch off properly
[02:53:54] <cradek> arcspiral.ngc :-)
[02:54:04] <jmkasunich> something along those lines
[02:54:09] <jmkasunich> plus the holes in the ears
[02:54:33] <jmkasunich> gotta add the drill and mill lengths to the tool table
[02:56:50] <jmkasunich> wow, some people are pretty dense
[02:57:08] <jmkasunich> somebody saw my Van Norman website and sent this "I will like to know if you sell the Van Norman "Ram Type" Milling Machines if yes.I will like to know the price and the type of payment."
[02:57:22] <jepler> hah
[02:59:47] <cradek> you could take his money and tell him you'll have one shipped to him from the past
[03:00:22] <jmkasunich> damn, I already sent a reply - I just remembered that HGR has (had) one recently
[03:01:46] <jmkasunich> yep, still do, $800
[03:15:43] <jmkasunich> was arcspiral generated by a program?
[03:16:37] <cradek> sure
[03:17:14] <jmkasunich> is it floating around somewhere?
[03:17:29] <cradek> I don't think so
[03:17:46] <cradek> it's very trivial though (if you use R format arcs)
[03:22:35] <jmkasunich> move the endpoint out and increase the R by the same amount, once per quadrant?
[03:22:59] <cradek> I think it was constant dtheta, pretty stupid
[03:23:48] <cradek> but yeah, step theta and take r=K*theta
[03:29:22] <CIA-22> EMC: 03seb 07TRUNK * 10emc2/src/hal/utils/bfload.c: Minor reorg for legibility, added some comments. No code changes.
[03:30:53] <CIA-22> EMC: 03seb 07TRUNK * 10emc2/src/hal/utils/bfload.c:
[03:30:53] <CIA-22> EMC: This adds a "list" command-line argument.
[03:30:53] <CIA-22> EMC: "bfload list" just probes for PCI cards and prints out what it finds.
[04:56:17] <jmkasunich> cradek: jepler: AXIS feature request - clear backplot when you hit the run button (if doing many of the same part, the backplot winds up with a lot of data in it)
[04:56:44] <jmkasunich> another feature request: a timer - cleared when you hit run, counts up minutes and seconds
[04:56:52] <jmkasunich> stops on end of program
[12:47:42] <jepler> skunkworks_: http://emergent.unpy.net/index.cgi-files/sandbox/demo_comp3.png
[12:50:16] <skunkworks_> that is awesome
[12:50:44] <skunkworks_> what is the left side showing?
[12:51:47] <jepler> my method works in two stages:
[12:52:22] <jepler> First, offset each arc or line by the given amount. If two adjacent moves intersect, then shorten them to the point of intersection. If they don't, join them with an arc.
[12:52:40] <jepler> Second, go through the path as a whole and loo for "loops" where the offset path intersects itself. Remove the loop.
[12:53:09] <jepler> So the funny stuff on the left side are the loops that are created by the simple method
[12:53:22] <skunkworks_> Ah - I think I understand. Very neat
[12:53:24] <jepler> the right side shows that my method correctly detects and gets rid of them, giving a nice offset path
[12:59:38] <skunkworks_> I can see how that is pretty much automatic pocket milling
[13:00:05] <skunkworks_> or could be.
[13:00:08] <jepler> yeah cradek also thinks it's close to being a general pocketing algorithm
[13:01:27] <fenn> what to do when there's a "c" shape?
[13:02:41] <fenn> will the algorithm leave a stub un-cut in the middle?
[13:02:54] <jepler> fenn: the present algorithm? yes, I think that's what it would do.
[13:03:29] <fenn> maybe there are some loops you shouldn't remove, like a double loop
[13:03:55] <fenn> i think that's what he meant by clockwise/counterclockwise loops
[13:04:05] <jepler> yes I think you're on the same track as cradek
[13:59:51] <jepler> fenn: here's an illustration of my present algorithm "losing" the part of the path that should be inside the circle: http://emergent.unpy.net/files/sandbox/lost-island.png
[17:03:57] <alex_joni> hi seb
[17:04:03] <seb_kuzminsky> hi :-)
[17:14:17] <alex_joni> seb_kuzminsky: what's new?
[17:14:23] <alex_joni> (besides the nice work on bfload?)
[17:15:11] <seb_kuzminsky> thanks, that's about it right now
[17:15:21] <seb_kuzminsky> once bfload is where i want it i'll return to my driver work
[17:15:48] <seb_kuzminsky> i've got something starting to shape up in a private revision control tree here, but it's too early to commit to our cvs
[17:16:21] <seb_kuzminsky> it currently can talk to hostmot2 on the 7i43 and the 5i20, but it only deals with gpios for now
[17:16:51] <seb_kuzminsky> the rest of the functionality should be straightforward to port from m7i43_hm2 (which is already in the tree but supports 7i43 only)
[17:17:27] <seb_kuzminsky> i think i just figured out the driver architecture needed to support dma and translation ram, so it should be pretty efficient when it's done :-)
[17:17:33] <seb_kuzminsky> what's new with you?
[17:18:19] <alex_joni> hmm.. been quite busy lately
[17:18:29] <alex_joni> not much time left for emc2 :(
[17:18:41] <seb_kuzminsky> real life intrudes eh?
[17:18:45] <alex_joni> started on cleaning up the code referring to the joints/axes issues
[17:18:57] <alex_joni> seb_kuzminsky: mostly work intrudes
[17:19:14] <alex_joni> just came back from germany today, will fly back on tuesday
[17:19:42] <seb_kuzminsky> where do you live?
[17:19:59] <alex_joni> timisoara/romania
[17:20:03] <seb_kuzminsky> cool
[17:20:14] <seb_kuzminsky> i used to live in sweden, but i live in the US now
[17:20:25] <alex_joni> oh, really?
[17:20:29] <alex_joni> were in sweden?
[17:20:52] <seb_kuzminsky> vasteras (outside stockholm) most recently
[17:21:08] <seb_kuzminsky> i was born in the backwoods of central sweden (parents were hippies there in the 70's)
[17:21:09] <alex_joni> I was in sweden 3 times so far
[17:21:19] <alex_joni> but only in the south.. near malmo
[17:21:29] <seb_kuzminsky> cool, i have family in lund near malmoe
[17:21:37] <seb_kuzminsky> nice place, i miss it sometimes
[17:21:44] <seb_kuzminsky> i've never been to romania tho
[17:21:44] <alex_joni> I actually was in lund 3 times
[17:21:57] <alex_joni> let me know if you plan to :P
[17:22:09] <seb_kuzminsky> ok we'll do a mini emc meet :-)
[17:22:12] <alex_joni> lund is great.. a very nice city
[17:23:50] <seb_kuzminsky> i'm converting my little benchtop mill to cnc...
[17:24:05] <seb_kuzminsky> i get all the computer stuff and most of the electronics, but i'm having a tough time with the mechanicals
[17:27:56] <alex_joni> well.. coding is not so easy on those ;)
[17:28:18] <alex_joni> * alex_joni just remembered he didn't read bofh in quite a while..
[17:34:57] <skunkworks_> alex_joni: which 2 packages do I need to install to get the real time kernel working on hardy? Is it the image and rtai 3?
[17:35:20] <skunkworks_> * skunkworks_ wishes he would have saved the conversation we had.
[17:35:46] <seb_kuzminsky> skunkworks: was it on irc? then it's probably been captured by the logger
[17:35:50] <seb_kuzminsky> logger_dev: bookmark
[17:35:50] <seb_kuzminsky> Just this once .. here's the log: http://www.linuxcnc.org/irc/irc.freenode.net:6667/emcdevel/2008-04-04.txt
[17:35:55] <skunkworks_> no - it was pm
[17:35:58] <seb_kuzminsky> ah
[17:35:59] <skunkworks_> :(
[17:36:24] <alex_joni> skunkworks_: I'm not sure if jeff moved them to linuxcnc.org
[17:36:27] <alex_joni> let me check
[17:36:42] <alex_joni> http://linuxcnc.org/ubuntu-hardy/base/binary-i386/
[17:36:44] <alex_joni> hmm, not yet
[17:36:57] <skunkworks_> heh
[17:37:20] <alex_joni> skunkworks_: they are still here to be found: http://dsplabs.cs.utt.ro/~juve/tempdebs/hardy/
[17:37:52] <alex_joni> you need the linux-image, the rtai-modules (the last one) and emc2_2.3.0~alpha..
[17:38:09] <skunkworks_> yes - I installed the linux-image and rtai-modules -3
[17:38:15] <seb_kuzminsky> alex_joni: how does rt latency on the new linux/rtai compare to the old one?
[17:38:17] <alex_joni> for additional drivers you might want to add linux-restricted-.. and linux-ubuntu-modules
[17:38:22] <skunkworks_> I am getting an error tryint to run the latency test
[17:38:28] <alex_joni> skunkworks_: which one?
[17:38:41] <skunkworks_> the one from the wiki - not latency-test
[17:38:47] <alex_joni> what error?
[17:38:48] <skunkworks_> let me pastbin it.
[17:38:51] <alex_joni> perfect
[17:39:04] <alex_joni> seb_kuzminsky: I didn't have dapper on my hardy machine
[17:39:17] <seb_kuzminsky> gutsy?
[17:39:20] <alex_joni> and my dapper machine is a bit older now, don't want to mess with it too much
[17:39:34] <alex_joni> no, nothing else, it's a new laptop
[17:42:40] <skunkworks__> http://pastebin.ca/971336
[17:43:22] <alex_joni> skunkworks_: dmesg?
[17:44:49] <skunkworks__> http://pastebin.ca/971341
[17:45:06] <skunkworks_> this is a core 2 duo
[17:46:16] <skunkworks_> it ran the livecd dapper fine.. around 16us
[17:47:26] <alex_joni> hmm.. feels like you don't have the correct packages installed
[17:47:37] <alex_joni> can you do dpkg -l | grep rtai ?
[17:47:52] <seb_kuzminsky> dpkg -l '*rtai*'
[17:48:15] <alex_joni> what's wrong with grep?
[17:48:26] <alex_joni> (din't know the * trick ..)
[17:48:30] <seb_kuzminsky> nothing
[17:48:39] <seb_kuzminsky> i just learned the '*' trick for dpkg -l myself & wanted to share
[17:48:44] <alex_joni> heh :D
[17:48:48] <alex_joni> show-off :P
[17:48:59] <seb_kuzminsky> * seb_kuzminsky blushes
[17:50:19] <skunkworks__> http://pastebin.ca/971345
[17:53:13] <alex_joni> skunkworks_: looks right :/
[17:53:41] <skunkworks_> anything else I can check?
[17:54:20] <alex_joni> check modinfo for /usr/real*/modules/rtai_hal.ko
[17:54:49] <alex_joni> skunkworks_: I assume you booted the right kernel..
[17:54:59] <skunkworks_> good question
[17:55:04] <alex_joni> hmm.. nope
[17:55:08] <alex_joni> there's ACPI all over it :P
[17:55:11] <alex_joni> check uname -a
[17:55:28] <seb_kuzminsky> skunkworks: looks like you booted -generic
[17:55:32] <seb_kuzminsky> (in dmesg)
[17:55:35] <alex_joni> dmesg
[17:55:35] <alex_joni> [ 0.000000] Linux version 2.6.24-14-generic
[17:55:38] <skunkworks__> heh
[17:55:50] <alex_joni> * alex_joni kicks skunkworks_ for giving him doubts
[17:56:12] <skunkworks__> I didn't pick a kernel - what ever it defaults to.. normally it is the 'right' on :)
[17:56:17] <skunkworks__> one
[17:56:23] <skunkworks__> biab
[17:56:34] <alex_joni> rtai is alfabetically after generic I think..
[17:56:48] <alex_joni> although it got before generic in my grub
[17:57:03] <seb_kuzminsky> you can tweak it in /boot/grub/menu.lst
[17:57:37] <seb_kuzminsky> i think by default grub should remember what you chose on the previous boot and use that
[17:58:22] <alex_joni> hmm, didn't see that ever
[17:58:30] <alex_joni> it just defaults to the first entry here..
[17:58:42] <seb_kuzminsky> i'm on 7.10
[17:58:44] <alex_joni> obviously you can arrange the entries how you like
[17:59:09] <alex_joni> C:\Documents and Settings\alex.joni\Desktop>uname -a
[17:59:10] <alex_joni> 'uname' is not recognized as an internal or external command,
[17:59:15] <alex_joni> * alex_joni grins
[17:59:23] <seb_kuzminsky> bbl, gotta go do the kind of work that makes money
[17:59:34] <alex_joni> well.. enjoy ;)
[18:00:03] <skunkworks_> sorry about that... That is the first time it didn't make the rtai kernel the top of the list.
[18:00:09] <skunkworks_> Working now.
[18:05:08] <alex_joni> skunkworks_: good numbers?
[18:08:40] <skunkworks_> na - 20us
[18:09:19] <skunkworks_> 21us
[18:17:17] <skunkworks_> it runs opengl really nice though
[18:24:09] <alex_joni> hmm, not that low.. but still ok
[18:24:19] <alex_joni> what kind of machine is this?
[18:26:39] <skunkworks_> its a 2.2ghz socket 775 core 2 duo. Onboard video (which doesn't seem to be causing problems) abit motherboard.
[18:27:26] <alex_joni> that's a E6400 ?
[18:28:28] <skunkworks_> e2200
[18:28:59] <skunkworks_> tigerdirect - whole system for under 200
[18:29:07] <skunkworks_> what do you expect :)
[18:29:41] <skunkworks_> dual core - sorry
[18:30:05] <alex_joni> sounds good
[18:31:01] <jepler> skunkworks_: that's 64-bit capable -- you should try out 64-bit ubuntu 8.04 beta and the 64-bit realtime packages
[18:31:14] <jepler> if you have nothing better to do with your time, that is :-P
[18:31:20] <alex_joni> oh, he doesn't :P
[18:31:47] <skunkworks_> what - you have some 64 bit packages? Location?
[18:32:04] <skunkworks_> * skunkworks_ goes and downloads the 64-bit beta
[18:32:08] <jepler> under http://linuxcnc.org/ubuntu-hardy/
[18:32:35] <jepler> look under base/binary-amd64 and emc2.3/binary-amd64
[18:33:07] <alex_joni> jepler: think you can put the i386 packages there too?
[18:33:45] <skunkworks_> * skunkworks_ has to wait 50 minutes
[18:36:34] <skunkworks_> so in the binary-amd64 - I just need linux-image and rtai-modules-2.6.24-..>
[18:37:16] <alex_joni> I think you need to install the 64 bi ubuntu too
[18:37:20] <alex_joni> bit..
[18:37:31] <alex_joni> (not just these 2 packages ontop of the 32bit OS)
[18:37:50] <jepler> alex_joni: the thing is that we have binary packages built from different sources but with the same version.
[18:38:10] <jepler> so we can't put them both in one tree
[18:39:23] <jepler> I think we need to resolve that by collaborating so that they both build from the same source packages (and higher numbers than are present in the repo now)
[18:39:28] <jepler> but I haven't taken the time to work on that yet
[18:40:55] <alex_joni> jepler: ok, I have some time this weekend, so call out if you want to do it
[18:41:27] <jepler> alex_joni: I'll be out of town
[18:42:42] <skunkworks_> alex_joni: I am downloading hardy 64
[18:44:28] <alex_joni> jepler: anything fun planned?
[18:44:51] <alex_joni> guess we'll do it some other time then..
[18:47:24] <jepler> alex_joni: it's not for work, dunno yet if it'll be fun
[18:47:54] <alex_joni> I'm flying to germany again next week (tue-thu)
[19:49:44] <skunkworks_> hmm - 64 bit hardy doesn't seem to see my sata hd - both dapper and hardy 32 did
[19:51:48] <alex_joni> maybe it's a 32bit hdd :D
[19:52:59] <skunkworks_> trying agian for good measure :)
[19:58:38] <alex_joni> * alex_joni heads to bed
[20:01:31] <skunkworks_> night alex
[20:01:40] <skunkworks_> It for some reason didn't see it the first time :)
[20:22:39] <skunk64> logger_dev: bookmark
[20:22:39] <skunk64> Just this once .. here's the log: http://www.linuxcnc.org/irc/irc.freenode.net:6667/emcdevel/2008-04-04.txt
[20:22:53] <alex_joni> wheee.. 64 skunks
[20:23:27] <skunk64> heh - painles
[20:23:30] <skunk64> painless
[20:24:16] <skunk64> 271mb updates
[20:24:24] <jepler> skunkworks_: ugh, patience is needed
[20:24:31] <skunk64> wow - fast servers
[20:24:38] <jepler> I might drop in during the weekend, but I'll be back for real tuesday
[20:24:41] <jepler> see you guys
[20:24:56] <skunk64> bye - have fun this weekend
[20:27:11] <alex_joni> see you jeff
[20:52:08] <skunkworks_> alex_joni: one thing.. I noticed that when you install the linux image - it doesn't ask for a reboot on the 32 bit version
[21:04:08] <skunk64> I installed linux-image-2.6.24-12-rtai_2.6.24-12.22.linuxcnc.1_amd64.deb and rtai-modules-2.6.24-12-rtai_3.6-~1_amd64.deb
[21:04:19] <skunk64> and emc2_2.3.0~alpha_amd64.deb
[21:04:45] <skunk64> but I cannot run either latency tests. Did I forget something?
[21:04:55] <skunk64> (I am booting off of the rtai kernel
[21:05:00] <skunk64> uname -a
[21:06:07] <skunk64> on the latency test I get Error opening /dev/rtf3
[21:07:01] <skunk64> and on the lateny-test I get http://pastebin.ca/971576
[21:08:58] <jepler> skunkworks_: oh yeah the rtai package is broken
[21:09:08] <jepler> it doesn't install a required package in /etc
[21:09:19] <skunkworks_> heh
[21:10:14] <jepler> I suspect there are other important bits missing too, but this is the first one .. put in /etc/udev/rules.d http://emergent.unpy.net/index.cgi-files/sandbox/99-rtai.rules
[21:12:49] <jepler> you'd also need these symlinks
[21:12:50] <jepler> 3255423 0 lrwxrwxrwx 1 root root 13 Mar 18 13:47 /usr/realtime-2.6.24-12-rtai/modules/rtai_ksched.ko -> rtai_sched.ko
[21:12:54] <jepler> 3255420 0 lrwxrwxrwx 1 root root 13 Mar 18 13:47 /usr/realtime-2.6.24-12-rtai/modules/rtai_mup.ko -> rtai_sched.ko
[21:12:57] <jepler> 3255421 0 lrwxrwxrwx 1 root root 13 Mar 18 13:47 /usr/realtime-2.6.24-12-rtai/modules/rtai_smp.ko -> rtai_sched.ko
[21:13:00] <jepler> 3255419 0 lrwxrwxrwx 1 root root 13 Mar 18 13:47 /usr/realtime-2.6.24-12-rtai/modules/rtai_up.ko -> rtai_sched.ko
[21:13:03] <jepler> 3255413 0 lrwxrwxrwx 1 root root 16 Mar 18 13:47 /usr/realtime-2.6.24-12-rtai/lib/liblxrt.so.1 -> liblxrt.so.1.0.0
[21:13:06] <jepler> 3255414 0 lrwxrwxrwx 1 root root 16 Mar 18 13:47 /usr/realtime-2.6.24-12-rtai/lib/liblxrt.so -> liblxrt.so.1.0.0
[21:13:09] <jepler> 3255351 0 lrwxrwxrwx 1 root root 10 Mar 18 13:47 /usr/realtime-2.6.24-12-rtai/include/asm -> asm-x86_64
[21:17:41] <skunk64> umm - what do I do with the symlinks?
[21:19:05] <seb_kuzminsky> you make them :P
[21:19:12] <seb_kuzminsky> ln -s Source Dest
[21:19:13] <seb_kuzminsky> just like cp
[21:19:24] <skunk64> * skunk64 isn't a linux person
[21:19:25] <jepler> in the named directory, you type e.g., ln -s rtai_sched.ko rtai_up.ko
[21:19:37] <jepler> or you wait a month for me to fix the deb and reupload it
[21:19:49] <skunk64> heh :)
[21:20:27] <skunk64> yes it took me that long to get the 99-rtai.rules into the rules.d directory
[21:21:01] <jepler> you might try without the symlinks, they probably only prevent building emc if they're missing
[21:21:14] <skunkworks_> heh - do I need to reboot?
[21:21:20] <jepler> I don't think so
[21:22:27] <skunk64> the rtai test seems to work.. Not the latency-test
[21:23:02] <skunk64> i might reboot
[21:23:48] <jepler> I think you actually need one of the symlinks .. this one: /usr/realtime-2.6.24-12-rtai/modules/rtai_ksched.ko -> rtai_sched.ko cd /usr/realtime*/modules; sudo ln -s rtai_sched.ko rtai_ksched.ko
[21:26:19] <skunkworks_> ok
[21:31:24] <skunk64> http://pastebin.ca/971618
[21:31:29] <skunk64> closer?
[21:33:41] <skunkworks_> I am leaving.. will read back. Again - have a good weekend.
[21:37:59] <seb_kuzminsky> skunkworks: apt-get install python-xml?
[21:45:09] <alex_joni> jepler: I believe I fixed that in the rtai source I built from
[21:45:16] <alex_joni> so if you build from that it might just be fine ;)
[21:52:18] <LawrenceG> ok... how do I find out what deb has tclConfig.sh ... it will be one of the -dev packages.... I am trying to build sim on 8.04
[21:55:28] <seb_kuzminsky> if you have auto-apt set up you can say "auto-apt search tclConfig.sh"
[21:56:00] <seb_kuzminsky> on my debian lenny/sid machine it's in tcl8.4-dev
[21:59:03] <LawrenceG> seb_kuzminsky, thanks... I installed auto-apt... that seems to only find files that are in installed packages... I'll see which tcl packages are on this box
[21:59:41] <seb_kuzminsky> LawrenceG: did you run auto-apt update? it should find all files in all packages, installed or not
[22:00:41] <LawrenceG> working...
[22:01:25] <seb_kuzminsky> yeah its slow...
[22:01:47] <LawrenceG> 14mb file
[22:02:44] <LawrenceG> and more....
[22:04:13] <LawrenceG> got it... wonder if I should build with tcl8.4 or 8.5
[22:06:25] <alex_joni> stick with 8.4
[22:06:32] <alex_joni> it's known to work..
[22:06:56] <LawrenceG> 8.4 seems to be the one installed on horny heron
[22:07:02] <seb_kuzminsky> use 8.5, it's 1.19% better ;-)
[22:12:05] <jepler> use 8.4. if you install 8.5, axis will crash at startup
[22:12:27] <LawrenceG> jepler, thanks... now onto BWidget
[22:12:30] <jepler> skunkworks: no idea about your last problem
[22:14:13] <alex_joni> jepler: isn't ext part of the xml parser?
[22:14:38] <alex_joni> LawrenceG: it's actually called horny hardon
[22:15:02] <jepler> alex_joni: could be -- maybe the package is python-xml or something?
[22:15:09] <jepler> really leaving this time. ..
[22:15:15] <alex_joni> yeah, I think I had to install that one
[22:15:37] <alex_joni> skunkworks: try installing the python-xml package if it's not already installed
[22:15:51] <alex_joni> jepler: don't forget your camera
[22:20:10] <alex_joni> * alex_joni is also leaving for bed this time..
[22:20:58] <seb_kuzminsky> later alex
[23:15:56] <jmkasunich> hi cradek
[23:16:02] <cradek> hi
[23:16:05] <jmkasunich> did you see my feature requests last night ;-)
[23:16:12] <jmkasunich> get busy!
[23:16:15] <cradek> nope
[23:16:19] <jmkasunich> dang
[23:16:37] <jmkasunich> cradek: jepler: AXIS feature request - clear backplot when you hit the run button (if doing many of the same part, the backplot winds up with a lot of data in it)
[23:16:41] <jmkasunich> another feature request: a timer - cleared when you hit run, counts up minutes and seconds
[23:17:11] <cradek> it's supposed to only keep so much. the limit may be too high (or maybe should be configurable)
[23:17:27] <cradek> jepler probably picked a reasonable number for his nice nvidia card
[23:17:55] <alex_joni> jmkasunich: want a link for the feature request trackers ?
[23:18:03] <cradek> hahaha
[23:18:12] <alex_joni> ;-)
[23:18:42] <jmkasunich> I just tossed those up last night when I thought of them
[23:18:59] <alex_joni> jmkasunich: I was only half-joking..
[23:19:19] <jmkasunich> I wanted to know how long my program was taking to run - I know predicting time ahead of time is non-trivial, but counting as the program runs is simple
[23:19:20] <cradek> clear before run might be nice, not sure
[23:19:25] <alex_joni> I think they are pertinent feature requests, but if you only mention them in here, they will probably be forgotten
[23:19:39] <jmkasunich> alex_joni: I know, and I'll try to remember to stick them up there
[23:19:49] <jmkasunich> last night it was late and I was pressed for time
[23:19:57] <cradek> for runtime I pretty much just look at the scrollbar that follows in the gcode
[23:20:08] <cradek> it's not time, but it sort of tells me how much is left
[23:20:47] <jmkasunich> even if there are loops?
[23:21:02] <cradek> no of course not :-)
[23:21:34] <jmkasunich> my usage is "ok, I ran this once, how long did it take, so I know how long it will take to run it 20 more times"
[23:21:54] <cradek> maybe that could go down in the status bar somewhere
[23:21:59] <cradek> I don't remember if there is room
[23:22:23] <jmkasunich> or be turned on/off like the DTG and velocity numbers in the DRO area?
[23:22:39] <cradek> yes, could go there easily enough
[23:23:55] <cradek> bbl
[23:24:07] <alex_joni> TTG ?
[23:24:21] <jmkasunich> no, elapsed time
[23:24:29] <alex_joni> hmm.. you could do both
[23:24:43] <alex_joni> but time to go would only be active after you ran the program once
[23:24:44] <jmkasunich> TTG is hard to do, especially with loops and probes and anything else that can change
[23:26:03] <alex_joni> right.. like FO