#linuxcnc-devel | Logs for 2015-03-25

[08:54:52] <skunkworks> 20um scalse probably are not enough either.
[08:55:08] <skunkworks> I should just try dual loop
[09:35:43] <skunkworks> kwallace, what was the final solution for the probing issue?
[09:36:02] <skunkworks> I see 1.6 is out.
[09:42:08] <kwallace> Hello skunkworks. So far, the solution is to drive the probe until it trips, record the position, drive a little farther (.015"), then retract. This allows the probe to be at a decent speed when the probe untriggers, leaving a shorter period where the dithering could happen. It works a lot better, but I still get an occasional error.
[09:43:25] <archivist> what sort of error
[09:46:06] <kwallace> I get something like "probe triggered while in non-probing MDI move". This happens during the G0 retract.
[09:48:31] <archivist> any idea of the contact material in the probe
[09:50:28] <archivist> I have an old probe here which seems to use stainless balls, a mistake as the reason it is stainless is the chrome oxide, an insulator
[09:53:29] <skunkworks> dad got 2 more probes of ebay for about $60... I don't know how he does it. Cat50 holders.
[09:54:02] <skunkworks> same brand he has now - some renishaw knock off.. works very well though
[09:54:13] <skunkworks> 'BIG'
[09:55:28] <archivist> I explored in mine http://www.collection.archivist.info/searchv13.php?searchstr=probe+2011
[09:57:24] <skunkworks> he has the one he got initally we are using on the k&t - then one that fits the matsurra http://electronicsam.com/images/KandT/probe/PROBE2.JPG
[09:57:26] <skunkworks> and now 2 more
[09:57:31] <skunkworks> archivist, neat
[09:59:17] <skunkworks> disassembly voids warranty...
[09:59:20] <skunkworks> ;)
[09:59:21] <archivist> the pressure needed to move the thing is too high so unlikely to let a sharp tool lean on it
[09:59:46] <archivist> warranty, wht the hell is that
[10:00:21] <skunkworks> still don't have a reciever...
[10:01:00] <pcw_home> kwallace: put probe signal through debounce on retract only?
[10:01:29] <archivist> I changed the LEDs on mine so I could see it and that is when I noticed dim leds due to the crap contact
[10:01:51] <skunkworks> High voltage!
[10:05:28] <archivist> then you will get that old arc crater problem, you must remember contact breakers on cars :)
[10:10:11] <pcw_home> 12 or 24V at a mA or so should not be harmful
[10:10:51] <pcw_home> (HV relative to 5V anyway)
[10:29:28] <Roguish> hey all. quick questions, please. where are the buildbot binarys kept? and how would I install one? and yes, I have been looking around a lot.... and can't find the binarys.
[10:31:33] <cradek> the front page at http://buildbot.linuxcnc.org/ tells you how to use the packages
[10:33:46] <Roguish> cradek: thanks, yes i see all that. but what if I want to install a previous version? i assume i could use 'apt-get install xxxxxxxx' but what the heck would the xxxxxx be?
[10:33:58] <Roguish> Oh, and that buildbot page is pretty good. i like it.
[10:34:25] <cradek> it kind of depends
[10:34:45] <cradek> a downgrade within one entry in sources, or across different ones?
[10:34:57] <cradek> you can say install linuxcnc=version (the = is an apt feature)
[10:35:14] <cradek> if you want to change sources, just remove --purge the old one and then change sources and install
[10:35:24] <Roguish> probably one entry (if by entry you mean like 2.6 vs 2.7)
[10:35:45] <cradek> brb
[10:59:35] <mozmck> hmm, I have twice now had firefox in linux here pop up a window with some resume builder ad - I wonder if I have a bug/virus?
[11:00:32] <cradek> more likely to be the website you were visiting at the time
[11:01:53] <mozmck> Well, it seems to have happened on two different ones now, and I think one was duckduckgo
[11:02:02] <cradek> hmmmm
[11:02:25] <mozmck> I'll have to pay better attention if it happens again. I assumed that was the case with the first incident.
[11:02:55] <mozmck> I have popup blocking turned on which seems to work pretty well, as well as adblock and ghostery installed.
[11:09:01] <Roguish> cradek: I am asking 'cause i had an unusual problem yesterday. I did a new install, lcnc worked fine. I did the update and lcnc quit. just would not run, in fact locked up the compute. I ran from a cmd window, and lcnc stopped at the 2nd line 'running server for TCP port....' just stopped and after a minute or so froze the computer. had to cold boot.
[11:10:05] <Roguish> so, I uninstalled with remove all. that was all with master.
[11:10:33] <Roguish> changed the repository to 2.6 stable and reinstalled. now works again.
[11:10:39] <cradek> see if you can figure out exactly how to reproduce it, and make a bug report please
[11:10:48] <cradek> I'm not familiar with that "TCP port" message
[11:11:29] <Roguish> i'll try. kinda FU'd my whole day yesterday. now I'm scared to update to master.
[11:11:43] <jepler> src/libnml/cms/tcp_srv.cc: "running server for TCP port %d (connection_socket = %d).\n",
[11:12:32] <Roguish> in the 'repositories', what happens if more than one 'linuxcnc' version is there? does it get confused?
[11:14:32] <jepler> Roguish: do you use nml-over-the-network? we are aware this is broken in 2.7 and master branches. For more information, see this commit message's full log and the discussions on the related SF bugs: c4c7fb9 nml: implement command queue with reliable reception
[11:14:50] <Roguish> I guess it's really 'branch' not version.
[11:15:26] <Roguish> nml over the network? like a Mesa ethernet board? no. simple 5i20 system.
[11:16:36] <jepler> no, nml-over-tcp is the feature that is supposed to enable user interface on one computer and linuxcnc on another computer
[11:17:29] <Roguish> not intentionally. guess i don't know for sure. is it something that must be intentionally turned on? or defaults on?
[11:17:54] <jepler> yes, it's not something you can enable accidentally
[11:17:58] <Roguish> is there a way to shut nml-over-tcp off?
[11:18:34] <jepler> it's only enabled when you put a modified emc.nml in your configuration files
[11:19:05] <jepler> (the file may actually still have to be named that, not sure)
[11:19:18] <Roguish> if it can be turned on accidentally, i may have done it, but haven't a clue what or how.
[11:19:51] <jepler> it seems unlikely then
[11:19:51] <Roguish> does the modified file come with an install/version/branch ?
[11:20:09] <Roguish> how does one tell the difference in the files?
[11:23:14] <jepler> I think it involves the inifile setting [EMC]NML_FILE
[11:23:20] <jepler> so if you don't have that in your inifile, then this is a red herring
[11:25:28] <jepler> like cradek, I'm not familiar with this message and I don't see it when running a sim/axis config at the tip of master branch: http://pastie.org/10053002
[11:26:16] <Roguish> I have 'debug' set to 77777777.
[11:26:46] <jepler> (time=1427299530.109748,pid=12269): Registering server on TCP port 5005.
[11:26:46] <jepler> (time=1427299530.109858,pid=12269): running server for TCP port 5005 (connection_socket = 4).
[11:26:53] <jepler> OK, with debug turned all the way up I do see that message too
[11:26:57] <Roguish> it would happen even when I tried to run the 'latency test' from the program menu.
[11:27:03] <jepler> the next message is typically
[11:27:04] <jepler> Starting LinuxCNC IO program: io
[11:30:12] <jepler> did you ever "sudo make install" a linuxcnc on this OS?
[11:30:33] <Roguish> no. totally new install from the hybrid.
[11:33:11] <jepler> between 2.6 and 2.7, I made a change to the way the process called "linuxcncsvr" starts up. If you somehow managed to mix a linuxcncsvr from an old version with a 2.7 or master version of linuxcnc, the startup process could hang at this point.
[11:34:00] <jepler> (in 2.6 and before, linuxcncsvr was unconditionally started in the background, followed by a 1s wait; in 2.7 and master, it is started in the foreground and automatically goes to the background after it is ready, eliminating the need for the 1s seleep -- a sleep which was not long enough on very slow systems anyway)
[11:34:18] <jepler> however, the most likely reason for that problem to crop up is if you ever did a "make install"
[11:35:27] <jepler> but as long as you install everything from a package, the debian packaging system dpkg will take care of removing the files from the other version before installing the files from the new version
[11:36:41] <Roguish> i generally use either the 'update manager' or 'synaptic' , only rarely will i use a cmd line apt-get.
[11:37:03] <jepler> they're using the same dpkg and apt-get behind the scenes, should be fine
[11:37:10] <Roguish> but i ask again, what happens if more than one branch is in the repository list?
[11:37:31] <Roguish> could something get confused or written over?
[11:37:56] <jepler> the system will prefer to install the highest version number
[11:38:29] <jepler> it is normal in a debian-based system for there to be multiple versions of a program installable. in the terminal, apt-cache policy packagename will tell you information about the versions available and installed for a given packagename
[11:38:47] <Roguish> I am looking for that SF bug you cited and can't find it. does it have an explicit bug number?
[11:38:48] * skunkworks is happy jeff is back...
[11:39:11] <Roguish> skunkworks: so is everyone!!
[11:40:25] <Roguish> does 'packagename' = 'linuxcnc' ???
[11:41:00] <jepler> yup
[11:42:08] <jepler> boy that's a long list of options available too
[11:42:25] <jepler> in my case, all the way from 1:2.6.0~pre4 to 1:2.8.0~pre1.568.ga73f462
[11:42:46] <jepler> so I've booted (a rather old) binary.hybrid.iso and I'm following the instructions to get master branch from buildbot..
[11:42:55] <jepler> it's now installing that 2.8.0~pre1 version
[11:43:20] <Roguish> well, hells bells. so many commands to know... wrote that on down for future use.
[11:43:44] <jepler> selected sim/axis configuration and it starts
[11:43:50] <Roguish> ok, how are you installing it? apt-get? what is the exact line, please?
[11:43:52] <jepler> so unfortunately I'm not reproducing your problem here
[11:44:02] <jepler> Roguish: I did:
[11:44:03] <jepler> sudo apt-key adv --keyserver hkp://keys.gnupg.net --recv-key E0EE663E
[11:44:17] <jepler> then put
[11:44:18] <jepler> deb http://buildbot.linuxcnc.org/ wheezy master-rt
[11:44:27] <jepler> in /etc/apt/sources.list.d/linuxcnc-buildbot.list using my favorite text editor
[11:44:30] <jepler> then sudo apt-get update
[11:44:33] <jepler> then sudo apt-get install linuxcnc
[11:45:18] <jepler> I did this all from an old (linuxcnc version 2.6.0~pre5) live CD image in qemu virtualization
[11:46:14] <Roguish> ok, so are you left with one version of linuxcnc only?
[11:46:26] <jepler> yes
[11:46:36] <cradek> apt will never install two of the same package
[11:47:19] <Roguish> I have a 'linuxcnc.list' but not a 'linuxcnc.buildbot.list'
[11:48:55] <Roguish> ok, i got the sequence. i generally use the updater, but it's the same under the hood, right? and i'll be sure to have only one branch in the repository list.
[11:49:33] <jepler> yes, I created the linuxcnc-buildbot.list file
[11:51:11] <Roguish> the updater uses nice little check boxes. same file(s) (I think) as synaptic uses.
[11:53:05] <Roguish> jepler, cradek: thanks for the help. I suspect there is a very obscure bug somewhere that I am good at finding. I will watch what I am doing and let you know if it happens again. and put in a but report, too.
[11:55:26] <cradek> our more recentish debian images have all those nice buildbot repos listed and commented out, which lets you enable or disable them with checkboxes in synaptic
[11:55:31] <jepler> OK. If you can give us steps how to reproduce it starting from the binary.hybrid.iso that'll let us diagnose what's going on
[11:57:22] <Roguish> i'll try if it strikes again. THANKS AGAIN. off to the shop to do some machining........
[11:58:20] <Roguish> p.s. I may try to find it this evening. it really pissed me off, and i like finding problems.
[11:58:26] <jepler> have fun, stay clean
[11:58:35] <jepler> well, have fun anyway
[20:29:34] <jepler> the_wench: tell dgarr you should take a look at the code here, I think it's a typo: tcl/tooledit.tcl:1104: switch $argct] {
[20:30:14] <jepler> hm do I not know how to work it?
[20:33:04] <Tom_itx> maybe add a ! before
[20:33:21] <Tom_itx> not sure about that bot
[20:33:27] <KGB-linuxcnc> 03Jeff Epler 052.6 5a0272d 06linuxcnc 10share/axis/tcl/axis.tcl axis: Use a preferred form of "switch" (closes: SF#411) * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=5a0272d
[20:33:54] <Tom_itx> !the_wench tell jepler this is how it works
[20:33:59] <Tom_itx> guess not
[20:35:24] <cradek> !later Tom_itx like this I think?
[20:35:25] <the_wench> will tell Tom_itx when he/she joins next
[20:42:19] <jepler> !later dgarr you should take a look at the code here, I think it's a typo: tcl/tooledit.tcl:1104: switch $argct] {
[20:42:19] <the_wench> will tell dgarr when he/she joins next
[20:42:25] <jepler> aha
[20:58:46] <Tom_itx> heh
[20:59:07] <the_wench> Tom_itx: cradek said like this I think?
[20:59:15] <Tom_itx> works
[21:55:32] <seb_kuzminsky> i changed wlo's robots.txt to allow w3c-checklink
[21:56:11] <seb_kuzminsky> to help us find broken links like Mark Johnsen reported
[22:09:56] <seb_kuzminsky> couple of broken links on this page: http://www.linuxcnc.org/index.php/english/weblinks/14-emc-related-pages
[22:10:00] <seb_kuzminsky> which i had never seen before
[22:15:57] <archivist> I should run my link checker I ran a few years ago on lunuxcnc
[22:21:40] <seb_kuzminsky> i used this one: http://validator.w3.org/
[22:22:36] <seb_kuzminsky> it's the web version of the checklink program we run on our asciidoc docs on every build
[22:27:37] <archivist> I did one over the main site and wiki back in 2012
[22:28:10] <archivist> cant remember what the name of the result page I made was
[23:02:19] <KGB-linuxcnc> 03Dewey Garrett 052.6 6c947aa 06linuxcnc 10tcl/tooledit.tcl tooledit.tcl: typo in switch * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=6c947aa
[23:02:20] <KGB-linuxcnc> 03Dewey Garrett 052.6 be6a81b 06linuxcnc 10src/hal/utils/halcmd_commands.c halcmd_commands.c: err msg applies pins or params * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=be6a81b
[23:03:02] <KGB-linuxcnc> 03Dewey Garrett 052.7 d6330e3 06linuxcnc 10share/axis/tcl/axis.tcl 10src/hal/utils/halcmd_commands.c Merge branch '2.6' into 2.7 * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=d6330e3