#linuxcnc-devel Logs

Oct 18 2017

02:04 PM Chris_Morley: Hi guys...I can't seem to push to linuxcnc anymore and i wrote an email to emc-developers about it and that was apparently rejected because i'm not subscribed? Could someone please look into that for me - Thanks
02:07 PM KimK: Hi Chris, good to talk to you. When you say you can't push to linuxcnc, I presume you know about the shutdown of git.linuxcnc.org and the accompanying move to Github?
02:11 PM KimK: Chris_Morley: Or have you been unsubscribed long enough that you missed that discussion? Anyway, maybe see https://github.com/LinuxCNC/
02:12 PM Chris_Morley: yes I got the emails about the change and i did change mine to pull from git hub but now I can't push
02:14 PM JT-Shop: did you change the push address?
02:14 PM Chris_Morley: I did what the email said: git config remote.origin.pushurl git@github.com:LinuxCNC/linuxcnc.git
02:15 PM KimK: Probably someone needs to click on something, give it a little time and someone will see your request here, I'm pretty sure.
02:15 PM Chris_Morley: The fact that the email list doesn't work either suggest to me something else is up
02:15 PM cradek: sourceforge demanded you click something a while back to remain subscribed
02:15 PM cradek: a lot of folks disregarded it or suspected it was fake
02:16 PM cradek: maybe look at the last few things you received?
02:16 PM Chris_Morley: oh hmm i've probably delete them... ok i'll check that out thank you
02:18 PM KimK: Also, I didn't change my git remote since I didn't have any work in progress, I just deleted and recloned from github.
02:21 PM Chris_Morley: ok I resubscribed see if that works... now the github thing....I do have ongoing work so don't wish to reclone.
02:23 PM KimK: OK, that's the right thing to do then.
02:25 PM cradek: Chris_Morley: when you try to push to github what do you see?
02:27 PM KimK: Oh wait, doesn't github use SSH keys?
02:28 PM cradek: yeah I was just wondering if cmorley's github account is set up right at all
02:28 PM cradek: I can see the account is c-morley and he has write access to linuxcnc
02:28 PM Chris_Morley: cradek: Permission denied (publickey).
02:28 PM Chris_Morley: fatal: Could not read from remote repository.
02:28 PM Chris_Morley: Please make sure you have the correct access rights
02:28 PM Chris_Morley: and the repository exists.
02:28 PM cradek: Permission denied (publickey)
02:28 PM cradek: yeah I think that means you need to set up a key for your github account
02:29 PM Chris_Morley: the key I had before with linuxcnc does not work?
02:29 PM KimK: It would work, but you have to tell github about it.
02:29 PM cradek: there's no way for anyone but you to enable a push key on github
02:30 PM cradek: looks like it's at https://github.com/settings/keys
02:30 PM cradek: after you sign in on the github website
02:30 PM KimK: github has a "paste block" for you to put your public key.
02:31 PM cradek: https://help.github.com/articles/connecting-to-github-with-ssh/
02:32 PM Chris_Morley: ok I will look into that - I don't remember any reference to having to do that in the email about changing to git hub - maybe I missed it. - thanks
02:33 PM cradek: a lot of folks were probably using github for other things, so they had done these steps already
02:34 PM cradek: aside from your own account key, I *think* everything is set up right for github c-morley but I can help if it's not
02:35 PM KimK: cradek: Ah, so *you're* the somebody who has to click on something, lol.
02:37 PM KimK: Well, thanks for doing that or the rest of us would be spinning our wheels.
02:41 PM cradek: welllll jepler and seb did more of the github magic than I did, but I do have some powers there that I try to use only for good, never for evil
02:44 PM Chris_Morley: Thank you -It seems I added a ssh key and it seems to work with dry-run. I used git hub with machinekit but I think I had to use a password each time - probably because I didn't use a ssh key. It's all pretty greek to me - lucily it doesn't change often :)
02:44 PM cradek: yay!
02:46 PM Chris_Morley: I don't know the difference between an id_rsa and an id_dsa but the dsa worked....
02:49 PM KimK: On an unrelated subject, I just returned from the WITS show http://www.witshow.org/ where I saw the 4'x4' version of this Oklahoma-built machine https://www.certmachines.com/joyeuse/ on display, running LinuxCNC!
02:55 PM KimK: I should add that Tormach (another LinuxCNC control) was not present, but I heard one exhibitor say that the show is continually sold out, and there is a waiting list of exhibitors who want to get in. They have to wait for someone to drop out.
03:34 PM cradek: if that's the one that's used for wheel jogging (where the target can move around willynilly) then I think yes
03:34 PM cradek: because somehow that actually seems to work right
04:02 PM KimK: That was my first thought after Seb said they were similar; is it possible to stimulate/irritate the trajectory planner in the same way, and see what happens? But maybe not needed?
04:23 PM andypugh: pcw_home: I have the fix for the hm2_dpll thing.
04:24 PM andypugh: But I also noticed that the docs say it only works for serial encoders.
04:24 PM andypugh: But it seems to be being applied to stepgens and quadrature encoders.
04:24 PM andypugh: The question is whether it actually wrks for these other things?
04:27 PM pcw_mesa: Yes it works for all. the docos are old
04:27 PM andypugh: I am a bit puzzled then, by this bit of code here: https://github.com/LinuxCNC/linuxcnc/blob/master/src/hal/drivers/mesa-hostmot2/dpll.c#L224
04:28 PM pcw_mesa: yeah thats weird
04:28 PM andypugh: I have a feeling it is never called.
04:28 PM pcw_mesa: i thought you just select the timer you wanted like you do with the incremental encoder or stepgen
04:29 PM pcw_mesa: are those just ?helpful? defaults?
04:29 PM andypugh: Looks likely
04:30 PM pcw_mesa: My instinct would be to dump them
04:31 PM andypugh: I think that’s right
04:31 PM andypugh: But I don’t have any absolute encoders to test with..
04:31 PM pcw_mesa: (after checking that you can indeed select the desired timer for each module in HAL)
04:32 PM andypugh: How many people do you think are using the absolute encoder module?
04:32 PM pcw_mesa: let me look at some abs encoder hal files
04:34 PM pcw_mesa: OK they seem to be set just like the quad encoder and stepgen (they have a timer pin) so defaults are not needed
04:36 PM andypugh: OK
04:37 PM andypugh: I will push the fix to 2.7 then delete the defaults from 2.8 when it upmerges
04:38 PM pcw_mesa: sounds right, thanks for looking into this, I think it was discovered by someone using absolute linear scales
04:39 PM pcw_mesa: they only used one timer but I think chose timer 2 for some reason ( which did not work )
04:40 PM andypugh: Took me a while to figure it out
04:40 PM pcw_mesa: i could not see it with a casual lookover
04:40 PM andypugh: It turns out that changes to timer 2 are only ignored if timer1 is positive.
04:41 PM andypugh: If timer1 is positive then it looks like FFFF1999 and the FFFF masks all the timer2 bits…
04:42 PM andypugh: Which was a bit of a “Doh!” moment.
04:53 PM andypugh: Why do you add encoders and stepgens to the dpll? Is it to reduce the effect of jitter? It sn’t to pre-fetch data like it is with the absolute encoders, I assume?
04:56 PM pcw_mesa: Yes on a a slow/jittery Ethernet host you may have 300 usec of position sampling jitter
04:56 PM pcw_mesa: this can be reduced to about 250 ns with the DPLL
04:58 PM pcw_mesa: ( though you may still have a slow +-10 usec or so baseline wander on a poor host dependent on machine loading )
05:01 PM pcw_mesa: the amount of pre-trigger needed depends on this baseline wander
05:06 PM pcw_mesa: Thank you!
05:09 PM andypugh: seb_kuzminsky: Would you care to opinionate on something? A recent bug-fix in 2.7 has a somewhat strained workaround to ensure that a silly thing that someone might have done still works.
05:11 PM andypugh: https://github.com/LinuxCNC/linuxcnc/commit/e472beab74e840498503a978b644cf6cc9fab05a#diff-7e8814606a607dce3608c2df3414d372R413
05:12 PM andypugh: I removed some (undocumented) default values, then put them back explictly if there are no addresses specified. I would prefer to remove this, for stylistic reaseons, in 2.8 where all configs get broken anyway…
05:14 PM andypugh: Actually, that reminds me, how do I merge the fixes in 2.7 up to master?
05:26 PM linuxcnc-build: build #123 of 4030.deb-stretch-rtpreempt-i386 is complete: Failure [4failed shell_3] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/4030.deb-stretch-rtpreempt-i386/builds/123 blamelist: Chris Morley <chrisinnanaimo@hotmail.com>, chris morley <chrisinnanaimo@hotmail.com>, John Thornton <bjt128@gmail.com>
05:56 PM linuxcnc-build: build #123 of 4031.deb-stretch-rtpreempt-amd64 is complete: Failure [4failed shell_3] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/4031.deb-stretch-rtpreempt-amd64/builds/123 blamelist: Chris Morley <chrisinnanaimo@hotmail.com>, chris morley <chrisinnanaimo@hotmail.com>, John Thornton <bjt128@gmail.com>
08:11 PM cradek: andypugh: git checkout master; git pull (just in case); git merge 2.7
08:17 PM andypugh: Then git push?
08:20 PM cradek: yeah
08:21 PM cradek: I like to check them both with git push --dry-run origin master 2.7, and if that result looks good I go ahead and push them both together
08:21 PM cradek: I like to look at the merge in gitk and make sure it makes sense (shows only what I expect)
08:23 PM andypugh: OK, tomorrow. Late here now.
08:24 PM cradek: g'night!