#linuxcnc-devel | Logs for 2016-05-01

[08:34:57] <pcw_home> For "eth-packet-loss", looks like sserial needs to be taught to cycle skip if theres a Ethernet RX packet loss/timeout:
[08:34:58] <pcw_home> http://www.pastebin.ca/3587778
[09:01:32] <jepler> PCW: so far there's no way for another part (sserial, encoder, etc) to know that a detected read or write error occurred, other than the fact that we now call all the force_write routines in either case
[09:01:51] <jepler> PCW: is this actually new with the branch and not in master?
[09:01:52] <jepler> hm2_eth: ERROR: used llio->read in realtime task (addr=0x5b00)
[09:02:40] <pcw_home> Let me check
[09:04:18] <pcw_home> only in eth-packet-loss
[09:08:13] <jepler> ok, noted
[09:22:53] <Roguish> pcw_home: hey Peter, how are those 5I20 boards I dropped off for repair? Had a chance to send in a batch yet?
[11:36:19] <skunkworks> 300ipm RUNS OK
[11:38:00] <skunkworks> adjusted the output scale so that the error was flat. then added just a pinch of d
[11:41:02] <skunkworks> Z I needed to lower P
[11:52:03] <KGB-linuxcnc> 03Dewey Garrett 05joints_axes13 de8a95e 06linuxcnc 10(8 files in 5 dirs) inihal: new ini hal pins for home, home_offset JA * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=de8a95e
[12:05:38] <Roguish_> would someone please explain what the joint-axes branch does, exactly? it looks interesting and i'ld like to try a sim in a vm.
[12:39:43] <jepler> Roguish_: linuxcnc has always given lip service to supporting non-trivial-kinematics machines, but JAx improves it. The name of the branch, "joints-axes" refers to one of the main elements of the work, clearly separating out, particularly in hal and inifile, what data pertains to a cartesian axis and what data pertains to a (possibly non-cartesian) joint
[12:53:30] <cradek> argh, my new machine communicates with the pluto fine, at least initally, but it's ferroring on many jogs
[12:53:34] <cradek> I bet it's getting read errors
[13:06:31] <cradek> yeah, pluto-servo.encoder.0.count changed from -2698 to +4 in one cycle (it was not moving)
[13:07:03] <cradek> 11 u32 RW 0x00000000 pluto-servo.communication-error
[13:09:18] <Roguish_> Jepler, thanks. that's pretty much what i thought. i'll plow through the docs and see if i can get a sim working in a vm. what dist would you suggest to start with?
[13:13:04] <jepler> Roguish_: ja will build and run on all the same distributions that linuxcnc master branch does.
[13:13:59] <jepler> cradek: that only gets incremented if there's an EPP timeout. It wouldn't tell you if just the data was bad
[13:14:10] <cradek> ah, thanks
[13:14:26] <cradek> I'll shoot in the dark and try another cable
[13:14:31] <cradek> it *almost* works
[13:14:58] <jepler> -2698 and +4 differ in a lot of bits
[13:14:59] <cradek> but -2698 to +4 doesn't seem like a single bit of error
[13:15:02] <cradek> heh
[13:20:14] <cradek> hm, when I wheelyjog Z it's always X that ferrors
[13:20:25] <cradek> if I do nothing but leave it in machine-on, it doesn't ferror
[13:20:42] <cradek> I think that means it's a noise thing
[13:36:53] <cradek> yeah a super long and slow XZ move is working just fine
[13:38:29] <cradek> man I hate when I lose gcode. I can remember cutting that on THIS machine but the gcode just isn't here
[13:39:18] <cradek> oh!! it's this .py right here
[13:39:44] <cradek> it was mathier than I remembered
[13:57:34] <jepler> http://bosch-work-memes.tumblr.com/
[14:29:01] <micges> jepler: around?
[14:33:42] <jepler> micges: in and out
[14:34:36] <micges> quick question
[14:35:29] <micges> say we have simple drill in preview
[14:36:08] <micges> but we have machine like morbidelli author 504 with 16 tools in one line
[14:36:19] <micges> and you can use up to 6 tools from that 16
[14:36:36] <micges> how can it be showed in preview?
[14:44:01] <jepler> ummm
[14:45:29] <jepler> good question
[14:46:00] <jepler> G4 dwell paints a little "X"
[14:46:05] <jepler> right at the controlled poin
[14:46:06] <jepler> t
[14:46:49] <jepler> perhaps you want to show some "X"es that represent the tools that are activated at any time, using their location relative to the controlled point
[14:48:34] <micges> I can try that
[14:48:36] <jepler> I don't know how, or if, it's customary to visualize this kind of part program
[14:49:07] <jepler> # self.dwells_append((self.lineno, self.colors['dwell'], x + self.offset_x, y + self.offset_y, z + self.offset_z, 0))
[14:49:56] <jepler> you'd just have to arrange to put the right coordinates in (the slightly misnamed for this purpose) self.dwells, in glcanon.py. I think the last number represents the plane
[14:50:38] <micges> coords are in tool table so it's easy to access them
[14:51:35] <micges> I can try X'es
[14:53:14] <micges> goal would be to add live plot of those additional tools but it's too messy it seems
[14:54:43] <micges> thanks
[14:59:24] <jepler> It sounds interesting. good luck.
[15:40:19] <skunkworks_> https://www.youtube.com/watch?v=81uwBAuR96I
[15:42:26] <Roguish> skunkworks: is that matsuura running its original motors and drives? or what?
[15:42:47] <Roguish> the company I'm at has one that looks exactly alike.
[15:49:19] <skunkworks_> yes - original motors/servos and spindle drive
[15:49:33] <Roguish> so just replace the brains....
[15:57:29] <skunkworks_> yes
[19:11:41] <seb_kuzminsky> i've got libbost-python-dev version, and libboost-everything-else 1.58.0+dfsg-5+b1
[19:11:46] <seb_kuzminsky> maybe that's got the bug jepler found
[19:12:07] <seb_kuzminsky> and i just havent updated the jessie buildbot vms recently, so they dont have this bug yet...
[19:23:50] <seb_kuzminsky> libboost-python1.60, also in jessie, doesn't have the problem
[19:32:22] <seb_kuzminsky> so just to be clear, it's a boost bug in 1.58, not a linuxcnc bug, false alarm
[19:46:25] <jepler> seb_kuzminsky: this seems vaguely familiar
[19:50:40] <jepler> ah yes https://sourceforge.net/p/emc/mailman/message/34690271/ ?
[19:56:58] <seb_kuzminsky> yeah
[19:57:19] <seb_kuzminsky> always fun when you google the error message and it leads you back to our own project
[20:14:29] <seb_kuzminsky> https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=823188
[21:02:46] <seb_kuzminsky> i said "jessie" above, but i meant stretch
[21:02:57] <seb_kuzminsky> i gonna go back to lurking