#linuxcnc-devel | Logs for 2014-05-21

[10:16:12] <kwallace2> Good morning. I'm trying to remember what component selects the closest pocket on a tool carousel -- something to do with entering a range, start number and target number and the output is the closest number. Any hints?
[10:21:02] <kwallace2> Am I back yet?
[10:22:02] <cradek> kwallace2: I did that in ladder - I'm not sure there's a specific component for it
[10:22:49] <kwallace2> I seem to recall it is a math function like mod?
[10:24:18] <kwallace2> I think I can find it on an old hard drive, if I boots up.
[10:24:22] <cradek> there's a modmath component
[10:24:58] <kwallace2> Okay, looking.
[10:25:15] <cradek> RTAPI_MP_INT(mod_dir, "Modulo direction blocks");
[10:25:35] <cradek> I see absolutely no documentation for it
[10:26:14] <cradek> halcmd: loadrt modmath mod_dir=1
[10:26:20] <kwallace2> http://wiki.linuxcnc.org/cgi-bin/wiki.pl?ContributedComponents#modmath_Modmath_functions_only_mod_dir_so_far
[10:26:20] <cradek> this gives me useful-looking pins
[10:26:46] <seb_kuzminsky> there's some documentation as a comment at the top of the modmath.comp file
[10:26:51] <cradek> there you have it
[10:27:11] <cradek> modmath.c
[10:28:12] <cradek> my ladder does auto-homing of the carousel, etc, too
[10:28:45] <cradek> the thought of doing nontrivial mathy or sequencing stuff with bare hal components makes me shiver
[10:30:11] <kwallace2> I haven't gotten into Ladder yet. I was hoping to avoid Python, but now I'm knee deep in the stuff. I rather like Python now.
[10:32:53] <kwallace2> Thanks for the help guys.
[10:37:56] <seb_kuzminsky> sure :-) want to transcribe the modmath.c comments into a modmath.9 manpage, to help the next guy?
[10:45:06] <kwallace2> Not finding documentation is why I submitted what I found to the wiki entry linked above, but then forgot about. You're right something more should be done, but man pages are another black hole for me. If I get a chance to figure it out I'll give it a try.
[10:52:41] <cradek> manpage markup is easy to understand if you are familiar with any markup-like language
[11:09:39] <kwallace2> cradek: I seem to recall manpage markup looking like awk, but it won't hurt to try.
[11:10:15] <seb_kuzminsky> i usually find a manpage that has roughly the same structure as what i want, and modify that
[11:12:09] <cradek> totally
[11:12:41] <seb_kuzminsky> copypaste > invent
[11:14:18] <kwallace2> Shoot, it takes me a half hour to look up the bits I forgot about in order edit a wiki page. (how to log in, layout, defeat CamelCase with <nowiki>, etcetera)
[11:15:32] <kwallace2> Life is rough, except the alternative.
[11:15:44] <seb_kuzminsky> and in the end you have a wiki page instead of normal documentation ;-)
[11:21:59] <kwallace2> What is normal?
[11:23:13] <archivist> the wiki pages go out of date, needs to be in main manual
[11:42:23] <micges> does anyone used this functionality? http://git.linuxcnc.org/gitweb?p=linuxcnc.git;a=commit;h=e5858b2f054a0cfaf6b773ad22d572d490e58322
[11:43:49] <micges> (some of ini settings are exported to hal and can be changed at runtime)
[11:47:45] <seb_kuzminsky> kwallace2: i consider manpages and asciidoc documentation normal
[11:48:33] <seb_kuzminsky> micges: i've never used it
[11:52:46] <kwallace2> "What is normal?" was jokingly meant in more of a general context :)
[12:00:17] <seb_kuzminsky> oops, i missed the joke :-/
[12:00:55] <seb_kuzminsky> i think the standard reply is: 'Normal' is a setting on my clothes dryer
[12:02:47] <archivist> I think striving to be normal is an illness :)
[12:25:04] <mozmck> striving to be different is it's twin brother :)
[13:40:29] <micges-dev> seb_kuzminsky: I found a typo in hal_lib.c: http://pastebin.com/0aTGwPSL
[13:40:47] <micges-dev> (gladly both sizes are the same)
[13:40:55] <micges-dev> seb_kuzminsky: ok for 2.6 ?
[14:06:41] <LAIR82_> Figured I might as well go straight to the top, I just pulled the latest and greatest, re-compiled and notice that when I go to run linux, the Create Desktop Shortcut link at the bottom of the selection page is grayed out. What could possibly make this happen?
[14:08:10] <cradek> perhaps http://sourceforge.net/p/emc/bugs/372/
[14:12:09] <skunkworks_> cradek: do we need a ok from rob before the new tp gets merged?
[14:12:28] <skunkworks_> (with your re-work on the git stuff?)
[14:16:35] <cradek> uhhhh
[14:16:37] <cradek> I dunno
[14:16:46] <cradek> I guess I was hoping he'd do it when he's ready
[14:17:01] <cradek> but I'm pretty sure he clearly said he was ready, so one of us could do it
[14:17:09] <cradek> what do you think?
[14:18:03] <skunkworks_> I am all for it. (and anyone can yell at me if they find an issue with it) ;)
[14:18:17] <skunkworks_> I am used to being yelled at.)
[14:20:20] <skunkworks_> one thing I just thought of.. rob changed the coloring of the blends.. (well I am color blind so he may have changed more - I don't know) parabolc blends are yellow. so you can see when the planner falls back to them..
[14:20:52] <skunkworks_> I think it is ok and my let people know if they don't have the ini setup.
[14:21:32] <cradek> is there any ini setup required, or does it all have sensible defaults?
[14:21:40] <cradek> (I feel strongly about there being sensible defaults)
[14:22:18] <skunkworks_> it runs parabolic blending (original tp) if now ini settings are present..
[14:22:37] <cradek> that's a bug we should fix
[14:22:53] <skunkworks_> (well with his refactoring and constriant violation fixes)
[14:23:53] <skunkworks_> you think the new behavour should be default? there are 4 settings with the new tp.
[14:24:42] <skunkworks_> let me find robs explaination
[14:24:44] <cradek> yes the improvements should be on by default! it's the whole point.
[14:26:57] <skunkworks_> http://linuxcnc.org/index.php/english/forum/10-advanced-configuration/27368-new-trajectory-planner-testersprograms-wanted?start=120#44419
[14:26:58] <skunkworks_> http://linuxcnc.org/index.php/english/forum/10-advanced-configuration/27368-new-trajectory-planner-testersprograms-wanted?start=130#44474
[14:27:05] <skunkworks_> actually 5 settings...
[14:27:33] <skunkworks_> the defaults he has there are what I have been using
[14:31:56] <skunkworks_> If I remember correctly - if arc blending is enabled - it has sane settings for the others. (with out specifying it in the ini) (but a look at the code would be in order)
[14:32:41] <skunkworks_> I am all for it.
[14:41:16] <seb_kuzminsky> micges-dev: that should probably go in 2.5
[14:41:35] <seb_kuzminsky> (the bug is present there too, and your proposed fix is clearly right)
[14:43:34] <micges> seb_kuzminsky: ok
[14:43:45] <seb_kuzminsky> skunkworks_, cradek: i had hoped to finish my new tp tests first, but it doesnt look like that's going to happen
[14:44:13] <seb_kuzminsky> so dont wait for me
[14:44:23] <seb_kuzminsky> i agree with cradek that it should be on by default
[14:44:31] <seb_kuzminsky> are the new ini settings documented?
[14:45:01] <skunkworks_> seb_kuzminsky: only on the 2 links I posted above.
[14:45:18] <skunkworks_> (pretty good explanations.)
[14:45:50] <cradek> forum posts are an even worse place for documentation than the wiki
[14:46:18] <cradek> micges: yes please do put that on v2.5_branch
[14:46:45] <skunkworks_> cradek: I agree...
[14:47:32] <skunkworks_> they should go somewhere.. I could put them on a wiki somewhere for now. (but it probably won't happen until I get back)
[14:47:55] <skunkworks_> http://electronicsam.com/images/house/DSC_0347.JPG
[14:47:59] <seb_kuzminsky> it's more fun to write code than to write docs, so i can understand procrastinating that part... but it *does* need to happen before release, and preferably before merge
[14:49:51] <seb_kuzminsky> skunkworks_: cute :-)
[14:52:16] <seb_kuzminsky> have any of you ever printed anything from ClassicLadder? me neither
[14:52:54] <skunkworks_> I tried it once.. I think there is an issue. IIRC Kim was trying to fix it...
[14:53:12] <seb_kuzminsky> the printing library that classicladder uses (libgnomeprintui) went away in Ubuntu Trusy 14.04
[14:53:33] <seb_kuzminsky> wow, it doesn't even work when the library *is* present... :-(
[14:53:55] <seb_kuzminsky> i'm going to remove libgnomeprintui-dev as a build dependency in 2.6 unless someone complains
[14:54:01] <seb_kuzminsky> in 10... 9...
[14:54:45] <cradek> I have probably done it once or more
[14:55:14] <cradek> it's fairly handy when working on nontrivial ladders
[14:56:08] <seb_kuzminsky> hmm, ok
[15:03:36] <skunkworks_> rough draft..
[15:03:45] <skunkworks_> http://pastebin.ca/2765738
[15:04:04] <skunkworks_> I will send it to rob and ask about the defaults and enabling by default.
[15:12:27] <KGB-linuxcnc> 03Michael Geszkiewicz 05v2.5_branch 7a76714 06linuxcnc 10src/hal/hal_lib.c hal: fix minor bug in allocating hal memory for HAL_S32 and HAL_U32 signals * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=7a76714
[17:46:43] <andypugh> micges: I have some unconnected encoders so sometimes see quadrature errors. I think you might have forgotten an \n in the HM2_ERR statements in d0bf76b1d47e4db7f9ea791fd13ccb12faaca1c8
[17:46:56] <andypugh> (the errors all wrap round for line after line)
[17:49:39] <andypugh> PCW: Has anyone reported problems with 7i73 encoders not counting? I can see 7i73 inputs 00 and 01 toggling as I move the wheel, the signal looks like quadrature, but the counts don’t count. I may be being bitten by a bug that I wrote..
[17:51:08] <seb_kuzminsky> i like it when my bugs bite me before they bite anyone else
[17:52:35] <andypugh> I just noticed that Aram upgraded LinuxCNC and now his 2.2 config won’t work. I bet that is my fault.
[17:52:40] <micges-dev> andypugh: thanks, checking..
[17:54:38] <micges-dev> andypugh: yep missing \n, thanks for report
[17:57:31] <andypugh> micges-dev: It possibly ought to not report the errors indefinitely
[18:18:38] <micges> it should report only one
[18:20:19] <micges> I tested it with hardware and it worked ok
[18:20:29] <micges> what is your setup?
[18:28:14] <andypugh> A 7i39 which generally has nothing plugged in (optional A-axis)
[18:30:07] <micges> and when second encoder error shows up?
[18:30:47] <PCW> 7I73 encoder counts work for me (at least enc0)
[18:30:48] <andypugh> I was getting a complete screen-full of quad errors
[18:31:18] <andypugh> But I actually have no idea what variant of LinuxCNC is on there at the moment.
[18:31:36] <andypugh> I think it might be a jog-while-paused JA4
[18:32:38] <micges> so it's before that fix of fix you mentioned
[18:33:38] <andypugh> Aye, I need to revert to something “normal” before getting carried away.
[18:33:54] <micges> or comment out this line
[18:34:10] <PCW> encoder errors really need an enable pin (so things like drives that generate encoder errors at startup can have the encoder errors masked)
[18:35:21] <PCW> also this would allow unused encoders to have any errors masked (in case you used encoder 0,1,3 for example)
[18:37:21] <micges> PCW: this error Andy mentioned have error enable pin, he just use code before that
[18:38:10] <PCW> OK I didn't notice that was there
[18:39:01] <PCW> I did check the A,B,I pins (halscoped them on a 7I77 which was not possible before)
[18:39:55] <micges> it works I assume? ;)
[18:41:12] <PCW> Yep though it was hard to turn the encoder slow enough to catch the index pulse at 1 KHz servo thread
[18:43:20] <micges> PCW: can you test it also on non muxed encoders?
[18:44:40] <micges> I don't have any setup for such tests atm
[18:46:05] <PCW> yeah I can check with a 7I76 config
[18:46:30] <PCW> (and not have to open the test computer box)
[18:57:35] <CaptHindsight> PCW: which mini-itx boards were you trying recently and did they have integrated LPT?
[18:59:19] <PCW> Gigabyte J1800 and it has built in LPT (that works in epp mode)
[19:00:09] <CaptHindsight> I need to slap it onto the back of an LCD
[19:00:20] <PCW> micges: A/B/I work on non muxed encoders also
[19:01:05] <PCW> maybe a thin mini-itx model (no PCI/PCIE expansion) would fit better
[19:02:11] <micges-dev> PCW: thats good, thanks
[19:02:15] <PCW> only problem with J1800s is that you need a new kernel if you want accelerated graphics
[19:02:30] <CaptHindsight> just run AXIS
[19:03:13] <PCW> even the funky old 2.6.32 RTAI kernel works (but without acceleration)
[19:03:21] <CaptHindsight> PCW: newer than the 3.4.55?
[19:03:32] <PCW> Yes 3.12 or >
[19:03:48] <PCW> Im running preemt-rt on mine
[19:04:12] <PCW> 2.6.32 RTAI was ~ 8 usec latencies
[19:04:43] <CaptHindsight> I just want to run 2 steppers off a tablet or touchscreen
[19:04:48] <PCW> plus about 3x as fast as Atom (NP with 5KHz servo thread)
[19:05:31] <CaptHindsight> hard to find anything with LPT or even an expresscard slot
[19:05:44] <PCW> etherhm2 and a cheap FPGA board
[19:05:45] <CaptHindsight> panel PC, all-in-one or laptop
[19:06:21] <CaptHindsight> is etherhm2 ready for prime time?
[19:06:52] <PCW> Well Ive had a stepgen config up 24/7 for a month now at 2 KHz without a hiccup
[19:07:08] <PCW> (on J1800)
[19:07:16] <PCW> so pretty close
[19:07:47] <CaptHindsight> the allwinner kernel and tool chain is still a magical train wreck
[19:07:55] <PCW> I just dont have a cheap ethernet endpoint (but working on one)
[19:08:21] <CaptHindsight> I need it by Tuesday :)
[19:08:45] <PCW> use a 7I76E (just huge and overkill)
[19:09:27] <CaptHindsight> maybe i should buy up every older keyboard PC with an LPT in sight
[19:09:36] <PCW> 5 stepgens 48 I/O , spindle encoder spindle 0-->10V RS422 expansion
[19:10:13] <PCW> Almost all the new J1800,J1900,J2900 MBs have LPT
[19:10:29] <PCW> many on backpanel DB25s
[19:11:30] <PCW> Gigabyte, Asus, ECS, Biostar, ASRock etc
[19:13:37] <CaptHindsight> then I still have to build it into an LCD+touch
[19:13:37] <PCW> a cheap one:
[19:13:39] <PCW> http://www.newegg.com/Product/Product.aspx?Item=N82E16813157498
[19:14:55] <PCW> well lay out a cheap FPGA card with a Micrel ethernet chip and I'll port the EtherHM2 firmware in a a day or two :-)
[19:16:05] <PCW> use a BBB
[19:16:22] <cradek> you already can't buy them
[19:16:35] <CaptHindsight> need to run a display
[19:16:55] <CaptHindsight> http://www.aliexpress.com/item/Useful-Industrial-Desktop-2-LPT-15-Inch-All-in-one-PC-QY-15C-DICA/918954436.html
[19:17:42] <KGB-linuxcnc> 03Michael Geszkiewicz 05araisrobo-scurve 194b61e 06linuxcnc 10(14 files in 5 dirs) tp: add support for reading JERK parameter from ini file and forward it to trajectory planner linear moves (no calculations yet) * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=194b61e
[19:20:25] <PCW> The ASRock MB has a GoodNight LED
[19:21:36] <PCW> speaking of which Dinnertime!
[19:21:38] <PCW> bbl
[19:22:54] <CaptHindsight> http://www.aliexpress.com/item/minipc-barebone-computers-with-6-COM-Parallel-LPT-Intel-Atom-D2550-dual-core-1-86Ghz-GMA3600/1689807229.html maybe this + LCD touchpanel
[19:36:02] <micges-dev> cradek: seb_kuzminsky: first results with araisrobo scurve code based on our master: http://ibin.co/1NESlMWW4jae
[20:56:12] <seb_kuzminsky> that's cool micges
[20:56:36] <andypugh> he left
[21:14:04] <cradek> I'd like to see the right end of it - it was the stopping that was the problem last time I looked at this code
[22:57:00] <CaptHindsight> PCW: "FPGA card with a Micrel ethernet" any favorite Micrel device or one/s to avoid?