Back
[00:02:16] <linuxcnc-build> build #2216 of 1405.rip-wheezy-armhf is complete: Failure [4failed compile] Build details are at
http://buildbot.linuxcnc.org/buildbot/builders/1405.rip-wheezy-armhf/builds/2216 blamelist: Dewey Garrett <dgarrett@panix.com>
[00:07:41] <linuxcnc-build> build #4036 of 0000.checkin is complete: Failure [4failed] Build details are at
http://buildbot.linuxcnc.org/buildbot/builders/0000.checkin/builds/4036 blamelist: Dewey Garrett <dgarrett@panix.com>
[07:28:30] <jepler> andy made hackaday
http://hackaday.com/2016/04/07/1921-ner-a-car-motorcycle-reborn-with-epic-parts-remanufacture/
[07:34:45] <jthornton> cool
[08:30:03] <skunkworks> I wonder if rtai acts differently
[08:46:28] <skunkworks> PCW, Could i have your gcode?
[08:46:38] <cradek> > It seemed pointless to make press tooling and only make two, so I made enough to satisfy worldwide demand for the next century.
[08:46:46] <cradek> I love andy's blog and his writing
[09:06:32] <pcw_home> http://freeby.mesanet.com/torus-test.ngc
[09:07:24] <pcw_home> (its just the torus.png image converted to gcode)
[09:28:49] <cradek> some good photos of the final result:
http://www.beautifullytold.com/ner-a-car/
[09:59:26] <skunkworks> pcw_home, 73k clock cycles on a 2ghz rtai so far (36us)
[09:59:59] <skunkworks> motion command handler tma
[10:00:04] <skunkworks> motion command handler tmax
[10:03:39] <lair82> Good Morning Guys, any idea what this error may mean?
https://app.box.com/s/rnjbd90ahq0ytq2m8zsi0m8ja8e7vgat
[10:04:35] <cradek> it expects CONFIG_DIR to be in your environment but it isn't
[10:07:48] <lair82> what would cause that to not be there, this machine has been running without any real issues, or errors like this, until yesterday, when all sorts of interesting things started happening.
[10:13:20] <JT-Shop> hard drive problem?
[10:15:37] <lair82> The PC on the machine completely shut itself down about a half hour ago, right in the middle of a running program
[10:16:27] <JT-Shop> motherboard dirty?
[10:17:01] <lair82> Last night, late, they dropped a program into the nc_files folder, remotely from the office programming computer, and it shut itself down
[10:17:21] <cradek> that's sure the smell of a hardware problem, isn't it
[10:17:48] <lair82> there is some dust on the fan, and the blades on the heat sink are a little dirty, but not oily or any thing
[10:18:33] <lair82> I have another MB and CPU spare on the shelf, do you guys think that may be my issue?
[10:33:17] <JT-Shop> check the health of the hard drive
[10:33:24] <JT-Shop> do a memory test
[11:52:37] <skunkworks> lowering the look ahead from 50 to 10 doesn't seem to effect it. peaking still at 70k clocks
[11:59:24] <skunkworks> I am sure it still does the same math for the path... the path just might be slower
[12:47:34] <seb_kuzminsky> thanks for checking
[12:52:39] <skunkworks> J1900 gigabyte
[12:53:13] <skunkworks> seems to stay under 80clocks with rtai
[12:59:36] <skunkworks> JT-Shop,
http://www.cnczone.com/forums/linuxcnc-formerly-emc2-/304280-forum.html
[12:59:39] <skunkworks> :)
[13:01:58] <pcw_home> math library issue?
[13:03:25] <skunkworks> I will reboot to rt_preempt to see if there is a difference
[13:03:58] <skunkworks> program is almost done
[13:05:36] <jepler> scripts/linuxcnc.in:CONFIG_DIR="${INIFILE%/*}"
[13:05:36] <jepler> scripts/linuxcnc.in:export CONFIG_DIR
[13:05:44] <jepler> CONFIG_DIR is set by the linuxcnc runscript
[13:08:45] * skunkworks thinks a hard drive failing
[13:20:19] <skunkworks> with rt_preempt has peaked motion-command-handler to 100k so far
[13:22:59] <skunkworks> although this is some odd rt_preempt kernel..
[13:23:58] <skunkworks> 3.2.0-4
[13:25:12] <lair82> Are there any precautions to take on mounting/connecting the SSD, it is only about a year and a half old, and it was new out of the box when installed,
http://www.newegg.com/Product/Product.aspx?Item=N82E16820721107&ignorebbr=1&cm_re=sv300s37a120g-_-20-721-107-_-Product
[13:27:33] <pcw_home> It that a drive from "anonymous"
[13:32:16] <skunkworks> I would go with samsung...
[13:32:26] <skunkworks> or crucial
[13:32:57] <lair82> Kingston junk?
[13:34:41] <skunkworks> I am not going to be that harsh.. But if I was doing something that I wanted the drive to last a long time I would go samsung pro series
[13:34:56] <skunkworks> or atleast an evo
[13:38:17] <lair82> ok, I will be ordering some shortly
[13:41:03] <skunkworks> This one running the torus is peaking at 108k clocks so far. I will let it run
[13:42:14] <skunkworks> (58us)
[13:46:51] <skunkworks> 2.7.4
[13:50:13] <PCW> but much lower on RTAI?
[14:06:06] <jepler> I think on x86 we directly assemble sin() cos() etc into inline fsin/fcos instructions. we certainly aren't using functions which have all the IEEE-correct handling that glibc's libm does...
[14:55:41] <skunkworks> zlog
[15:03:25] <skunkworks> pcw_home: seems higher with rt_preempt - when I left it was up to 118k clock cycles
[20:24:08] <jepler> hmm. half-baked idea. X[1e5] is not ambiguous in the same way X1E5 is. I wonder if making number-inside-brackets parse different than bare number is easy or hard.
[20:24:46] <jepler> but I think it would be best to put a fix in that prevents e-format numbers from ever being substituted in gladevcp or whatever the culprit is, and we already have some snippets that do that so forget I said anything
[20:50:31] <jepler> >>> "%f" % 1e100
[20:50:31] <jepler> '10000000000000000159028911097599180468360808563945281389781327557747838772170381060813469985856815104.000000'
[20:50:38] <jepler> of course just using %f has its own share of problems
[22:50:39] <KGB-linuxcnc> 03Dewey Garrett 05joints_axes13 73af6b8 06linuxcnc 10src/emc/usr_intf/axis/scripts/axis.py axis.py use same value as iniaxis.cc:loadAxis() JA * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=73af6b8
[22:50:39] <KGB-linuxcnc> 03Dewey Garrett 05joints_axes13 e183b4c 06linuxcnc 10lib/hallib/basic_sim.tcl 10lib/hallib/sim_lib.tcl sim_lib.tcl improve compat with non-iden kins JA * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=e183b4c
[22:50:39] <KGB-linuxcnc> 03Dewey Garrett 05joints_axes13 537b388 06linuxcnc 10tcl/tklinuxcnc.tcl tklinuxcnc: joint/world sw for non-iden kins JA * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=537b388