#linuxcnc-devel Logs

Apr 05 2017

#linuxcnc-devel Calendar

06:22 AM jthornton: I'm getting a couple of errors trying to build mint 18 deb http://paste.ubuntu.com/24319661/
06:23 AM jthornton: I've tried to install the unmet build dependencies but they are not found... must be missing an important clue...
06:39 AM jthornton: looks like /debian/configure needs a line added after case $DISTRIB_NAME in for LinuxMint-18
08:08 AM jthornton: down to python-support not found
08:09 AM jthornton: the RIP seems to work fine, so I wonder if I need python-support for the deb?
09:03 AM jepler: jthornton: master branch seems to have a stanza in debian/configure for LinuxMint-18. if you are looking at packaging 2.7 then study that first.
09:04 AM jepler: if you are trying to package master branch then I dunno, there's supposed to be support for it but maybe it's broken
09:04 AM jepler: anyway you'll run into other problems with 2.7 on newer linux distros, like the AXIS y-axis jog bug that is due to a python-tkinter bug that we didn't incorporate a workaround for in 2.7
09:04 AM JT-Shop: jepler: yes, I was trying for 2.7
09:04 AM jepler: and in master branch it just happens to be fixed by an unrelated change...
09:04 AM jepler: so anyway have fun and afk again
09:05 AM JT-Shop: thanks, I may skip 2.7 and just do master
10:36 AM -!- #linuxcnc-devel mode set to +v by ChanServ
10:54 AM -!- #linuxcnc-devel mode set to +v by ChanServ
11:22 AM seb_kuzminsky: good morning
11:59 AM JT-Shop: morning
12:59 PM stustev: Has there been any progress on the handwheel miscount I described on the email list? If not, what do I need to do now?
01:38 PM seb_kuzminsky: stustev: remind me, what's your scale and gain?
01:44 PM seb_kuzminsky: i think jmk is right, as usual, and it's a floating point rounding problem
01:46 PM pcw_mesa: getting rounding "right" on signed numbers is often a bit fussy
01:46 PM seb_kuzminsky: we have round(3) on all supported platforms
01:47 PM seb_kuzminsky: ilowpass uses floating point math inside, before going to s32 on its hal pins
01:55 PM pcw_mesa: Hmm IEE754 default is bankers rule
01:56 PM pcw_mesa: so round(3) does not use the defualt
01:57 PM seb_kuzminsky: hmm
01:58 PM seb_kuzminsky: i think in this case it won't matter, because we should be much closer to the correct integer than one half
01:59 PM pcw_mesa: does ilowpass use round for the final conversion (I would guess it does)
01:59 PM seb_kuzminsky: no, it just casts from double to s64 to int to s32
02:04 PM stustev: scale is 1.0 - gain is 0.005
02:05 PM seb_kuzminsky: i reproduce the issue here when i use low gain
02:05 PM stustev: ok - what is a reasonable gain?
02:06 PM seb_kuzminsky: the problem goes away for me if i raise gain above 0.5, which is a suspicious number
02:06 PM seb_kuzminsky: but failures with low gain is definitely a bug
02:08 PM stustev: ssssoooo - I will play with the gain to see what happens. I am not on a physical machine so I will not be able to report axis motion characteristics
02:17 PM seb_kuzminsky: i modified ilowpass to change its internal float counter to a hal pin, so we can see what it's thinking:
02:17 PM seb_kuzminsky: http://highlab.com/~seb/linuxcnc/ilowpass-problem.png
02:17 PM seb_kuzminsky: that's .scale=1, .gain=0.1
02:17 PM stustev: problem goes away here with gain of 0.5
02:18 PM stustev: will try 0.1
02:18 PM seb_kuzminsky: it's weird, because the gain is only used to set the internal value, which looks fine to me
02:18 PM seb_kuzminsky: it's not used to set the output pin from the internal value
02:19 PM seb_kuzminsky: that's done with the rounding/truncating implicit in casting from float to int
02:22 PM stustev: .498 shows the error - .5 is good
02:31 PM seb_kuzminsky: halscope really doesn't like it when you unload the component that its scoping
02:42 PM skunkworks: well - that wasn't well thought out.. ;)
02:43 PM stustev: in interp_convert.cc - version 2.7 - line 1745 - should 42.2 be 42.1?
02:50 PM KGB-linuxcnc: 03Chris Radek 052.7 bf4f128 06linuxcnc 10src/emc/rs274ngc/interp_convert.cc Fix error message typo found by stustev * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=bf4f128
02:53 PM stustev: version 2.8 line 1800 has g42.2
02:54 PM KGB-linuxcnc: 03Sebastian Kuzminsky 052.7-ilowpass-fix 9c4e934 06linuxcnc 10(6 files in 2 dirs) tests: reorg ilowpass test so i can add a low-gain test next to it * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=9c4e934
02:54 PM KGB-linuxcnc: 03Sebastian Kuzminsky 052.7-ilowpass-fix 1c6a07f 06linuxcnc 03tests/ilowpass/low-gain/README 03tests/ilowpass/low-gain/expected 03tests/ilowpass/low-gain/test.hal tests: add a test of ilowpass with low gain * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=1c6a07f
02:54 PM KGB-linuxcnc: 03Sebastian Kuzminsky 052.7-ilowpass-fix 95bd447 06linuxcnc 10src/hal/components/ilowpass.comp ilowpass: round the output instead of truncating * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=95bd447
02:54 PM seb_kuzminsky: stustev: try that branch
02:54 PM seb_kuzminsky: it's got the fix jmk suggested
02:55 PM stustev: will just a git pull get the fix to try?
02:55 PM seb_kuzminsky: you have to fetch, then 'git checkout 2.7-ilowpass-fix' and build
02:56 PM stustev: ugh - now I have to learn something - :)
02:56 PM stustev: will this be in a different directory - I have never done this
02:56 PM seb_kuzminsky: heh
02:57 PM seb_kuzminsky: no, by default it'll replace whatever you have in your git checkout
02:57 PM stustev: if I checkout in my current directory will it change to ilowpass-fix and then back with a different checkout command?
02:58 PM stustev: will try it - can't be any worse than starting over
02:59 PM KimK_laptop: you could also clone new to linuxcnc-devvvvv(?) and checkout there.
02:59 PM seb_kuzminsky: you need 2 more v's, but other than that it's right
02:59 PM KimK_laptop: heh
02:59 PM seb_kuzminsky: ;-)
02:59 PM seb_kuzminsky: bbl
03:04 PM JoeHildreth: Hello all. First time here and have a question about a manpage I have written for the hal_manualtoolchange user component. Who can I give this to to get into the documentation?
03:05 PM JoeHildreth: I think that standard procedure is for me to install git and pull the project, add my file then commit. I have never done that before.
03:06 PM stustev: now the count is correct on the positive side of zero and is behind by one on the negative side of zero
03:07 PM cradek: JoeHildreth: welcome! you're right, the best way is for you to make a git commit, and share that using some method, there are several choices
03:07 PM cradek: JoeHildreth: have you seen http://linuxcnc.org/docs/html/code/contributing-to-linuxcnc.html#_overview_of_the_process
03:07 PM stustev: the action is different in that the direction change does not correct the count as before - the count is always correct on the positive and always off by one on the negative side
03:08 PM cradek: pcw_mesa> getting rounding "right" on signed numbers is often a bit fussy
03:08 PM cradek: heh, pcw sure predicted that
03:09 PM KimK_laptop: Hi JoeHildreth, thanks very much for writing a manpage, documentation is always needed and appreciated.
03:10 PM cradek: JoeHildreth: it's a lot to learn to submit one change, but if we are lucky enough to have you continue to help the project in the future, it's time well spent
03:11 PM JoeHildreth: cradek: Yes, I have read the docs on that page. I guess I just need to get over the intimidation of the process. My usage of git has only been on my local machine for writing components for Joomla CMS and needless to say my skills are a bit limited on using it.
03:11 PM cradek: JoeHildreth: lots of folks here can help
03:13 PM cradek: JoeHildreth: the most popular way right now is probably to use a github pull request
03:13 PM JoeHildreth: So let me ask a dumb question, I just pull the master branch, add my hal_manualtoolchange.1 file to the linuxcnc/docs/man/man1/ folder and then commit?
03:14 PM cradek: yes, and when you commit be sure to do the signoff (git commit -s)
03:14 PM cradek: check your commit to see that your name and email are right in it, and it contains what you want
03:15 PM cradek: then you can make a pull request on github, OR git format-patch -1 and send it to the developers' list
03:15 PM stustev: interp_convert.cc in 2.8 has the same 42.2 in it that 2.7 had - line 1800 - does that need to be fixed or will it get fixed automatically?
03:16 PM cradek: stustev: a future merge will fix it, let me try it now
03:17 PM cradek: JoeHildreth: about 8.3 in the docs, a missing manpage is enough like a bug that you could choose to put your new documentation in the 2.7 branch instead of master. as far as I'm concerned, either is fine.
03:18 PM KGB-linuxcnc: 03Chris Radek 05master c291cbd 06linuxcnc 10src/emc/rs274ngc/interp_convert.cc Merge branch '2.7' * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=c291cbd
03:18 PM JoeHildreth: cradek: If I pull 2.7 branch, it will end up in the master branch too?
03:19 PM cradek: yes, we use the "merging up" workflow, so if you put something in 2.7 it will get merged into master, see section 4.3 in the contributing docs describing it
03:20 PM cradek: actually that's what I just did with stustev's fix, we put it in 2.7 (although he didn't give me a git commit, he just explained the change in english) and then merged it up into master
03:20 PM cradek: the main drawback of him not making a commit is he doesn't get recorded as the author of the change
03:32 PM JoeHildreth: cradek: I think i screwed up. the I filled out a pull request, but that would be for the completed code I sent correct?
03:32 PM JoeHildreth: I need to download the code, add my changes, commit then a pull request...
03:33 PM cradek: > WJHildreth wants to merge 1,410 commits into 2.7
03:33 PM cradek: yes this is wrong :-)
03:34 PM cradek: I'm not sure what you did, but the pull request should show just your change
03:35 PM cradek: I closed it for you
03:35 PM cradek: but in retrospect I think you could have done that
03:36 PM cradek: (unfortunately I'm sure not a github expert)
03:36 PM JoeHildreth: cradek: I am sorry. I just cloned the source.
03:38 PM JoeHildreth: cradek: OK, I just copied my manpage to the linuxcnc/docs/man/man1 folder.
04:52 PM seb_kuzminsky: stustev: bummer
05:20 PM stustev: cradek: my explanation of the change was in the form of a question - I was not sure a change was needed :)
06:18 PM KGB-linuxcnc: 03Sebastian Kuzminsky 052.7-ilowpass-fix 1b35d6c 06linuxcnc 10tests/ilowpass/low-gain/expected 10tests/ilowpass/low-gain/test.hal tests: test negative numbers in ilowpass/low-gain test * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=1b35d6c
06:18 PM KGB-linuxcnc: 03Sebastian Kuzminsky 052.7-ilowpass-fix e11fa93 06linuxcnc 10src/hal/components/ilowpass.comp ilowpass: handle negative numbers correctly * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=e11fa93
06:19 PM seb_kuzminsky: maybe that?
06:19 PM seb_kuzminsky: stustev: that version does better on negative numbers
06:46 PM stustev: yes - this works better - I tried .5, .1 and .005 - all count correctly - thanks
06:46 PM stustev: be back in the morning
07:16 PM linuxcnc-build: build #4864 of 1300.rip-precise-i386 is complete: Failure [4failed compile runtests] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/1300.rip-precise-i386/builds/4864 blamelist: Chris Radek <chris@timeguy.com>, John Thornton <bjt128@gmail.com>
07:17 PM linuxcnc-build: build #4075 of 1301.rip-precise-rtai-i386 is complete: Failure [4failed compile runtests] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/1301.rip-precise-rtai-i386/builds/4075 blamelist: Chris Radek <chris@timeguy.com>, John Thornton <bjt128@gmail.com>
08:01 PM linuxcnc-build: build #4882 of 0000.checkin is complete: Failure [4failed] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/0000.checkin/builds/4882 blamelist: Chris Radek <chris@timeguy.com>, John Thornton <bjt128@gmail.com>
08:55 PM seb_kuzminsky: sigh, task: main loop took 35.128954 seconds
09:43 PM JoeHildreth_: I have been trying to figure out the github stuff so that I could add the manpage I wrote for the hal_manualtoolchange component. These are the steps I have done and I think I am good. Could someone confirm.
09:44 PM cradek: that looks perfect! let me merge it for you
09:45 PM JoeHildreth_: I forked the code, pulled to my machine, added the file, done a commit and signed it, pushed back to my repo, then done a pull request. The pull request says that it is running the travis cl build.
09:45 PM JoeHildreth_: This right?
09:46 PM JoeHildreth_: I was on the master branch, so can you get it to the 2.7 branch?
09:46 PM cradek: if you want it on both, because of the mergeup workflow, you'd want to make the pull request against 2.7
09:47 PM cradek: although I can also move it for you
09:50 PM cradek: also our policy is to use unobfuscated email addresses, as it says in #7 in the contributing doc
09:50 PM cradek: sorry this first time is a bunch of things to figure out
09:50 PM cradek: I can fix any or all of these for you, if you don't want to mess with it
09:50 PM JoeHildreth_: If I would have forked the 2.7 tree, would it have automatically moved to the master 2.8 branch too?
09:51 PM cradek: yes if your commit was on top of the 2.7 branch, it would "merge up" into master
09:51 PM JoeHildreth_: Yes, if you would do that for me. That would be awesome.
09:52 PM JoeHildreth_: OK, well I think I would like to help with documentation where I can. I will delete the 2.8 fork that I have now and fork the 2.7 branch for future additions.
09:52 PM cradek: in your commit that you already have, you probably just have to do: git checkout 2.7
09:53 PM JoeHildreth_: Thanks for being so patient with me. I have never done this before and trying to wrap my shriveled brain around it.
09:53 PM cradek: it's totally worth helping when someone wants to help!
09:54 PM JoeHildreth_: Yes, that command switched me to 2.7. Thanks! That saves me some time.
09:54 PM JoeHildreth_: I still have loads to learn. I will likely have several questions along the way.
09:56 PM JoeHildreth_: I see now that all checks have passed. Now it has to be merged by one of the project members, correct? Nothing left for me to do on this one?
09:56 PM cradek: right
09:57 PM cradek: I'm moving it to 2.7, fixing your email, and will push it for you shortly
09:58 PM JoeHildreth_: So the docs said to obfuscate my email. Should I change it so that it is correct?
09:59 PM cradek: eek, where did you see that?
10:01 PM KGB-linuxcnc: 03Joe Hildreth 052.7 3b6c6ba 06linuxcnc 03docs/man/man1/hal_manualtoolchange.1 Add hal_manualtoolchange manpage. * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=3b6c6ba
10:01 PM KGB-linuxcnc: 03Chris Radek 05master a38685a 06linuxcnc Merge branch '2.7' * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=a38685a
10:01 PM cradek: there it is! thank you!
10:01 PM JoeHildreth_: http://linuxcnc.org/docs/devel/html/code/contributing-to-linuxcnc.html
10:02 PM cradek: I think you misread? it says unobfuscated.
10:02 PM JoeHildreth_: DOH! I am sorry, I misread that. I will fix and promise not to do that again.
10:02 PM cradek: no problem, fixed!
10:06 PM JoeHildreth_: Will it eventually show on the documentation part of the linuxcnc website, or is there somewhere else in the code tree that I should be looking at?
10:07 PM cradek: the docs on the website build themselves, so I think it'll show up in a while
10:07 PM cradek: at the top docs page http://linuxcnc.org/docs/html/ you can click "Expand Man Pages" and see them all
10:08 PM cradek: if there's another step to make yours show up .... I don't know what it is, but jthornton would
10:09 PM cradek: if you look at the rightmost column of http://buildbot.linuxcnc.org/buildbot/grid you'll see your commit is currently building
10:09 PM cradek: shortly I think you'll see the 2000.docs step happen, and that will cause the update
10:11 PM JoeHildreth_: So I can identify my stuff by the SHA1 key at the top? Sorry for the neophyte questions.
10:12 PM cradek: yes every commit gets a unique hash like that
10:12 PM cradek: and if you click the upper right "building" (0000.checkin row) you'll see it's yours
10:16 PM JoeHildreth_: Neat! Thanks for the information and help. Are there any resources you would recommend that I read/watch to get over the learning curve?
10:17 PM cradek: we try hard to have that be the contributing document, but it's hard to know exactly what to put in there.
10:18 PM cradek: we want it to be super helpful with linuxcnc-specific knowledge, but not be a unix tutorial or git tutorial or rehash things that are lots of places elsewhere on the net
10:19 PM JoeHildreth_: I understand.
10:19 PM cradek: like it should probably at least mention buildbot I suppose? but it shouldn't be a github tutorial because that's something github should be better at doing
10:19 PM cradek: I bet your thoughts about the document, considering your fresh eyes, would be helpful
10:20 PM JoeHildreth_: I think I will get the github and git part figured out as there are a lot of resources out there.
10:21 PM JoeHildreth_: I have used LinuxCNC for a few years now, but I am at the point that I really want to have a better understanding of how it works. Maybe find a way that I can help as a way of saying thank you.
10:21 PM cradek: git is a truly great tool with a truly terrible user interface, so lots of people have written help pages :-/
10:21 PM cradek: that's super, I appreciate it, as a guy who's worked on it for lotsa years
10:22 PM JoeHildreth_: Most of my coding has been on a local devel environment as the only coder, so the whole git thing will take some time for me to learn.
10:23 PM JoeHildreth_: I do have some pretty good resources available for that.
10:23 PM cradek: bbl, hungry
10:23 PM cradek: thanks JoeHildreth
10:24 PM JoeHildreth_: I think the real stumbling block for me was the forking thing. I cloned the source from the linuxcnc repo but I cannot push to it. The docs should mention that, perhaps?
10:24 PM JoeHildreth_: Thanks. G'Night.
10:24 PM DrippityDrops: I wish there was an easy way to help the community more, but I don't know how to put my work in a place for people to screen it.
10:25 PM DrippityDrops: I have work that is usable for people but I don't even know what to call these things.
10:25 PM JoeHildreth_: What do you mean?
10:25 PM DrippityDrops: there needs to be an online sandbox which is a linux build that people can add things to
10:26 PM DrippityDrops: and it be.... a simulation of sorts
10:26 PM JoeHildreth_: You mean to test configurations or add software?
10:27 PM DrippityDrops: just like when you run linuxcnc and choose a sample .hal with simulated hardware. it would be like that but this huge list of everyones contributions as some usable interface you can try the work out on
10:28 PM DrippityDrops: well for example, if a person added new kinematics, or an unusual interface for some special job, you could upload it to the server and people could try it out and download the :tools: they need for their own build
10:29 PM JoeHildreth_: I see. That would be cool.
10:29 PM DrippityDrops: like glade panels, hal examples for unusual card sorts
10:30 PM DrippityDrops: just like a big sandbox so people can make adjustments and revisions to the works in an atmosphere they are used to working in. just a place to BUILD A CONFIG
10:30 PM JoeHildreth_: That is sort of where I am, I want to know how it all fits together. Looking at sample files and reading tons of misc docs has helped. I think it would be nice to have a YouTube
10:31 PM JoeHildreth_: series that really talks about how the bits work together, how they are configured. Many people learn better by video presentation and example.
10:31 PM DrippityDrops: im quite the opposite far too impatient for most videos
10:31 PM DrippityDrops: unless the person is very very informative and quick moving
10:32 PM DrippityDrops: The biggest problem with linuxcnc I think is the userbase is too small to get lots of very sharp people who can ALSO contribute
10:33 PM JoeHildreth_: Unfortunately, I think some of the stuff could get bogged down. So your idea would be some interactive simulator that you can see what is happening in the RT system as you make adjustments to the configs?
10:34 PM DrippityDrops: it would be like a sample configurations list you that all users can upload to
10:34 PM DrippityDrops: and download from
10:34 PM DrippityDrops: easy to page through on why each build is special
10:34 PM JoeHildreth_: Other than docs and maybe some video screen casts and that sort of stuff, I may not be of much help to folks. All my coding is web related and no python experience at all.
10:36 PM JoeHildreth_: Hey, take that one step further and have drag and drop configs. OK, I have a stepper system, so I drag the stepper icon and it adds the stepgen and all the other bits after collecting a few bits of information.
10:37 PM JoeHildreth_: Next I grab the probe and get the info for the probe, then the home switches, then the VFD, etc.
10:38 PM JoeHildreth_: Well, I gotta run. Have a great night.
10:38 PM DrippityDrops: take care