#linuxcnc-devel Logs

Jul 01 2017

#linuxcnc-devel Calendar

08:21 AM jthornton: trying to install master uspace on stretch and after adding the repository and doing sudo apt-get update I get a GPG error http://paste.ubuntu.com/24996251/
08:38 AM skunkworks: ran linuxcnc over night on stretch running a 7i92 - no realtime errors on this laptot
08:47 AM jepler: you're running on a teapot?
08:47 AM jepler: (nice, thanks for the report)
09:19 AM skunkworks: heh
09:39 AM jepler: jthornton: your log shows you: sudo apt-key adv --keyserver hkp://keys.gnupg.net --recv-key 3cb9fd148f374fef
09:40 AM jepler: W: GPG error: http://buildbot.linuxcnc.org stretch Release: The following signatures couldn't be verified because the public key is not available: NO_PUBKEY EF1B07FEE0EE663E
09:40 AM jepler: but the key in this other step has a different key ID
09:40 AM jepler: if you want to trust that other key ID you need to add it too
09:40 AM jepler: 3cb9fd148f374fef vs EF1B07FEE0EE663E
09:45 AM jepler: (it is expected that the buildbot and wlo have different keys)
09:51 AM JT-Shop: thanks
10:17 AM jthornton: what is the url for the wlo? I'm assuming now that the repository needs to point to it...
10:19 AM jthornton: be back this afternoon
10:39 AM seb_kuzminsky: jthornton: deb http://linuxcnc.org stretch base
10:50 AM KGB-linuxcnc: 03Norbert Schechner 052.7 89d8101 06linuxcnc 10src/emc/usr_intf/gmoccapy/gmoccapy.py 10src/emc/usr_intf/gmoccapy/release_notes.txt gmoccapy_1_5_8 - added Num_Pad jogging * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=89d8101
11:05 AM KGB-linuxcnc: 03Norbert Schechner 05master 9b487e5 06linuxcnc 10src/emc/usr_intf/gmoccapy/gmoccapy.py 10src/emc/usr_intf/gmoccapy/release_notes.txt gmoccapy_2_3_1_1 - added KeyPad Jogging * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=9b487e5
11:13 AM KGB-linuxcnc: 03Norbert Schechner 05master e21752a 06linuxcnc 10docs/src/gui/gmoccapy.txt docs - gmoccapy - added KeyPad Jogging informationn * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=e21752a
11:17 AM KGB-linuxcnc: 03Norbert Schechner 052.7 d3715ed 06linuxcnc 10docs/src/gui/gmoccapy.txt docs - gmoccapy - added info about NumPad Jogging * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=d3715ed
11:19 AM KGB-linuxcnc: 03Norbert Schechner 05master 16e7b99 06linuxcnc 10docs/src/gui/gmoccapy.txt docs - gmoccapy - added description from KeaPad to NumPad * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=16e7b99
11:55 AM jepler: I added preseed support to live-wrapper, new images will take advantage of that
12:15 PM seb_kuzminsky: sweet
12:16 PM seb_kuzminsky: i just tried to boot my bridgeport off the amd64-r4 image, forgetting that it's not a 64 bit cpu...
12:24 PM jepler: does it give a crummy, crummy error for that condition?
12:26 PM jepler: actually it gives a good error. yay.
12:27 PM seb_kuzminsky: yeah, it sure clued me in
12:33 PM jepler: r5 will be up in about 5 minutes, has improvements like "group dialout", no prompt for root password on install, and tweaks to the list of packages (glxgears installed now, halcompile verified working)
12:36 PM jepler: .. up
12:51 PM seb_kuzminsky: with r4, my atheros "AR5212/5213//2414" pci wifi card worked without any fuss
12:53 PM seb_kuzminsky: there was a video mode problem at boot
12:54 PM seb_kuzminsky: http://i.imgur.com/Xut2hVi.jpg
12:54 PM seb_kuzminsky: i picked 303 and it went into the graphical installer
12:54 PM seb_kuzminsky: iirc there was trouble with the linuxcnc-wheezy iso's video mode picker too
12:55 PM seb_kuzminsky: i dont remember if cradek solved it or if i just picked a video mode off the list or what
12:55 PM seb_kuzminsky: bbl
01:22 PM jepler: I don't see where he did anything
01:29 PM jepler: this is interesting. it seems to work. https://github.com/jepler/stretch-live-build#experimental-persistence
02:36 PM seb_kuzminsky: jepler: could you add gparted to the live system?
02:36 PM seb_kuzminsky: the installer can resize partitions but not move them...
02:37 PM seb_kuzminsky: i don't know why Les Newell's patches dont apply for me :-(
02:39 PM jepler: does someone know why mousepad has historically been considered an unsatisfactory gui text editor?
02:39 PM hazzy: It would be really nice to have gparted on the live boot!
02:39 PM jepler: I was looking at the one in stretch, and it has multiple tabs, syntax highlighting, ... not sure what else to look for
02:40 PM seb_kuzminsky: vi key bindings?
02:40 PM seb_kuzminsky: sorry, i'll show myself out
02:41 PM jepler: seb_kuzminsky: (I'm a vi user personally so I get what you're saying)
02:42 PM jepler: (vim-gtk3 is going to be in the next image as well)
02:42 PM seb_kuzminsky: how much would vim-nox add?
02:43 PM seb_kuzminsky: Installed-Size: 2682
02:43 PM seb_kuzminsky: i'll bribe you with beer next time you come to visit...
02:45 PM jepler: seb_kuzminsky: 10.3 vs 10.6 according to apt install
02:46 PM jepler: it pulls in e.g., the ruby interpreter even though eew
02:46 PM seb_kuzminsky: hazzy: i booted to the live system (on -r4), joined my local wifi network, and apt-get install'ed gparted, it worked fine
02:47 PM seb_kuzminsky: but the network's not always there, and its nice not to need it
02:50 PM jepler: $ git am ~/sshfs/afk/home/jepler/0001-Added-PIN_NAMES-option-to-mb2hal.patch
02:50 PM jepler: Patch format detection failed.
02:50 PM hazzy: seb_kuzminsky: I have done that to, many times! Thats why it would be nice to have it in the live :)
02:50 PM jepler: seb_kuzminsky: looks like a well-meaning^Wbroken mailer escaped "From " at the beginning of a line with ">"
02:56 PM seb_kuzminsky: i see that you're right
02:56 PM seb_kuzminsky: thanks for figuring it out
02:58 PM KGB-linuxcnc: 03Les 05master ba45d09 06linuxcnc 10src/hal/user_comps/mb2hal/mb2hal.h 10src/hal/user_comps/mb2hal/mb2hal_HOWTO.ini 10src/hal/user_comps/mb2hal/mb2hal_hal.c 10src/hal/user_comps/mb2hal/mb2hal_init.c Added PIN_NAMES option to mb2hal * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=ba45d09
02:58 PM KGB-linuxcnc: 03Les 05master f53ece1 06linuxcnc 10src/hal/user_comps/mb2hal/mb2hal.c Fixed excessive CPU load in mb2hal * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=f53ece1
03:09 PM jepler: .. OK, gparted and vim-gtk3 are in. I'll wait for more feedback before rolling out a -r6
04:37 PM skunkworks: hmm - why isn't my postion feedback showing off in the axis display... I am having a following error of .5 inches - and command vs actual is the same.
04:38 PM skunkworks: seems feedback is hooked up net motor.00.pos-fb <= hm2_[HOSTMOT2](BOARD).0.stepgen.00.position-fb axis.0.motor-pos-fb pid.0.feedback
04:38 PM skunkworks: this is pcw's config
04:43 PM skunkworks: just the fact that I have axis.0.f-error - wouldn't that discrepancy show up in the preview?
04:43 PM skunkworks: when set to actual position..
04:59 PM jepler: skunkworks: Sounds weird. is this on stretch or something older? 2.7? backlash or screw comp? homed?
04:59 PM skunkworks: stable (2.7) on stretch
04:59 PM skunkworks: yes homed. no screw comp or backlash
05:01 PM skunkworks: I could be I am missing something in the hal file.. (I thought maybe commanded was hooked to fb or something. But I am not seeing that.
05:02 PM skunkworks: https://pastebin.com/f0iiTybz
05:03 PM skunkworks: (I certainly could be missing something)
05:03 PM jepler: is motor-pos-fb right in hal?
05:05 PM skunkworks: net motor.00.pos-fb <= hm2_[HOSTMOT2](BOARD).0.stepgen.00.position-fb axis.0.motor-pos-fb pid.0.feedback
05:05 PM skunkworks: seems right
05:07 PM skunkworks: stepgen feedback is hooked to axis.0.motor-pos-fb
05:07 PM skunkworks: and again - I am seeing following error - that is a motion thing - isn't it?
05:08 PM skunkworks: that would mean commanded vs actual was different.
05:09 PM seb_kuzminsky: jepler: i instelled your r4 on my bridgeport and it homed and i jogged it around. looks good!
05:09 PM skunkworks: yay!
05:09 PM seb_kuzminsky: i know you said you already fixed the dialout group thing, which was the only thing i had to tweak (to talk to my vfd)
05:10 PM seb_kuzminsky: yeah it's pretty great :-)
05:10 PM seb_kuzminsky: one strange thing happened that i didnt debug...
05:10 PM skunkworks: cutting edge - really
05:10 PM seb_kuzminsky: i logged in, played around, noticed the dialout problem, added myself, logged out, and...
05:10 PM seb_kuzminsky: when i logged back in the panels didnt show up
05:11 PM seb_kuzminsky: it's cool that i could use the same exact linuxcnc ini and hal files on stretch/uspace/2.7 as what i've been using on wheezy/rtai/2.7 for ages
05:11 PM seb_kuzminsky: no changes needed, nice work in the uspace thing too
05:12 PM skunkworks: amazing
05:16 PM skunkworks: if I unhook axis.0.motor-pos-fb
05:17 PM skunkworks: the cone still moves if I have axis interface viewing commanded and actual...
05:18 PM skunkworks: although the cone is more jerky when viewing actual...
05:19 PM seb_kuzminsky: i'm taking off for an extended 4th of july family "camping" trip, i'll be back around thursday
05:19 PM seb_kuzminsky: until then...
05:19 PM Tom_itx: be safe
05:21 PM skunkworks: have fun!
05:26 PM skunkworks: so my thinking is this. I have a following error of 2 inches at the moment.. (program paused)
05:27 PM skunkworks: I should be able to switch between the commanded position and actual position in the axis interface menu and see the cone jump
05:27 PM jepler: skunkworks: it seems like!
05:28 PM skunkworks: ok.. then maybe I am not crazy
05:32 PM jepler: my first little sanity check doesn't show a behavior like yours: http://paste.debian.net/974289/
05:32 PM jepler: but at this point when it has f-errored, cmd and fb are equal again, because in *machine off* fb is copied to cmd...
05:33 PM jepler: if I understand you right, your linuxcnc *hasn't* ferrored, right?
05:34 PM jepler: halcmd: setp ini.0.min_ferror 1000
05:34 PM jepler: halcmd: setp ini.0.ferror 1000
05:35 PM jepler: once I do this, I can jog with the feedback disconnected, and it does "chunk", but pressing @ shows two different numbers for X
05:35 PM jepler: 21 float OUT 3.228183 axis.0.motor-pos-cmd ==> Xpos
05:35 PM jepler: 21 float IN -1 axis.0.motor-pos-fb
05:35 PM skunkworks: correct - f-errror is large here for testing
05:36 PM jepler: your UI is axis ?
05:36 PM skunkworks: yes
05:36 PM jepler: and you're looking at the DRO numbers?
05:36 PM jepler: oh you mention the cone at some point. the cone always follows commanded, I think
05:36 PM skunkworks: dro numbers don't change either
05:37 PM skunkworks: (switching between commanded and actual)
05:37 PM jepler: 21 float OUT 1.650739 axis.0.joint-pos-cmd
05:37 PM jepler: 21 float OUT -2.0125 axis.0.joint-pos-fb
05:37 PM jepler: do joint -cmd and -fb differ?
05:38 PM skunkworks: No - odd
05:38 PM skunkworks: where is the f-error coming from then?
05:38 PM skunkworks: hold on - I seemed to have stopped the program
05:39 PM skunkworks: ok - I see the dro change.
05:40 PM jepler: ok
05:40 PM skunkworks: darn - i thought the cone would follow commanded vs actual.
05:40 PM jepler: yes fbofw the cone follows commanded and I think it always has
05:40 PM skunkworks: because it does show actual when you are moving with power off
05:40 PM jepler: I no longer remember the motivation and I'd be willing to revisit the question but probably not during stable 2.7.x
05:40 PM skunkworks: (with encoders)
05:41 PM jepler: that's because feedback is copied to command with machine off (stepper or encoder)
05:41 PM skunkworks: ok
05:41 PM skunkworks: so - I could setup a logging computer if I want to show this... ;)
05:43 PM jepler: are you deliberately creating a large ferror?
05:45 PM skunkworks: yes - I was trying to show how not having enough stepgen headroom causes f-error
05:45 PM jepler: ah
05:46 PM skunkworks: halscope shows it - but say you have the following error set high - and you don't have enough velocity headroom - you can mill quite a ways off from where you think you are.
05:46 PM skunkworks: so the backplot would be a good tool
05:47 PM jepler: that's true
05:47 PM skunkworks: some people have a following error and then just set the f-error larger and don't fix the problem
05:49 PM skunkworks: well - it is a corner case.. I haven't thought I needed it until now ;)
05:58 PM skunkworks: jepler, would that be a bug report or feature request?
06:00 PM jepler: skunkworks: feature request
06:00 PM jepler: not that github really has a distinction, they're all Issues
06:02 PM skunkworks: ok
06:02 PM jepler: https://emergent.unpythonic.net/files/sandbox/0001-UNTESTED-unconditionally-show-feedback-position-in-b.patch
06:05 PM skunkworks: jepler, ooh... Thank you! Now I need to see if I can build rip on stretch :)
06:07 PM jepler: skunkworks: haha my trick has worked
06:12 PM skunkworks: well - :0
06:18 PM jepler: so it seems that if you have a user called "0day", and you configure a systemd "service" with [Service]User=0day, the service runs as root instead https://github.com/systemd/systemd/issues/6237
06:18 PM jepler: as usual the public face of systemd leaps in to yell at the user for configuring their system stupidly
06:18 PM skunkworks: 0day sounds like a virus
06:19 PM jepler: https://www.youtube.com/watch?v=rjigODNy3jk
06:57 PM skunkworks: cool - build-deps works!
07:03 PM skunkworks: slow INTERNET here. 14% so far of build-deps
07:03 PM jepler: boo
07:03 PM skunkworks: I don't think build deps worked when I was using jessie initially - I remember getting them all by hand
07:03 PM jepler: you should specify --no-install-recommends if yo uare manually installing build-deps, or just use mk-build-deps -irs sudo
07:04 PM jepler: -irs is unrelated to the IRS
07:04 PM skunkworks: heh
07:04 PM skunkworks: as long as I don't mute the speakers - or turn on the keyboard light - latency on this laptop is good ;)
07:07 PM jepler: I've been mixing a new drink lately: 2oz gin, 1oz lime juice, honey or simple syrup to taste. Stir in ice, strain into glass. Top with 4oz of your favorite IPA. Based on this "franco-american 60" recipe http://www.kindredcocktails.com/cocktail/franco-american-60
07:09 PM skunkworks: hmm - sound interesting
08:03 PM skunkworks: building
08:04 PM skunkworks: lots of warnings?
08:07 PM skunkworks: *a small handful of warnings...
08:09 PM skunkworks: builds and runs.. Now to apply the patch. bbl
08:09 PM skunkworks: Yay!
08:30 PM seb_kuzminsky: jepler: i just tried to install the i386-r4 iso on an amd64 machine, the installer environment ran and it rebooted to the grub menu fine, but when i asked it to boot the new stretch install it just rebooted
09:55 PM jepler: seb_kuzminsky: boo. I wonder why.
09:55 PM jepler: I have only installed in VMs but I have done 32-bit (BIOS) and 64-bit (BIOS and non-secure UEFI) there
09:55 PM jepler: maybe not with r4, though(?)
09:56 PM jepler: you get the grub menu but rather than booting into the kernel next, it just reboots? or you never get the grub menu?
10:22 PM seb_kuzminsky: i get the grub menu, then i select the top item, then it thinks for a little bit, then it reboots and i get back to the grub menu
10:23 PM jepler: that's weird
10:25 PM jepler: removing "quiet" from the kernel commandline might let you glimpse something before it goes all wrong. you could also try the rescue mode. beyond standard troubleshooting suggestions, I don't know what data to ask you for
10:30 PM seb_kuzminsky: i tried rescue mode, same result
10:30 PM seb_kuzminsky: i should try !quiet
10:30 PM jepler: huh rescue mode is useless on a working system, it just tells you the root account is locked
10:30 PM jepler: rescue mode implies !quiet
10:31 PM jepler: my other suggestion is, try with the debian netinst image or debian live image and see if it's stretch or if it's my image
10:31 PM jepler: afk and goodnight
10:32 PM seb_kuzminsky: g'night
10:33 PM seb_kuzminsky: i think i'm going to go try to rescue the hackspace shopbot tonight, instead of debugging this issue :-/
10:33 PM seb_kuzminsky: user reports: i commanded 1200 rpm and got 9196 rpm
10:33 PM seb_kuzminsky: with the new hy_gt_vfd driver i just thought i finished debugging