#linuxcnc-devel Logs

Jul 21 2017

#linuxcnc-devel Calendar

09:49 AM skunkworks: is there a how-to for switching from rtai to uspace on the wheezy livecd?
09:57 AM skunkworks: someone went from r1 2.7.9 apt-get update/upgrade to 2.7.10 and says the computer slowed way down. is that possible?
09:58 AM cradek: I think there's not enough information to say
09:58 AM cradek: what packages upgraded? what is slow now?
10:04 AM skunkworks: it is on a d525 (kinda suprised it ran 'fast' in the first place)
10:04 AM skunkworks: https://forum.linuxcnc.org/9-installing-linuxcnc/33046-how-to-downgrade
10:04 AM skunkworks: (we talked by one another a few times.. :) )
10:05 AM cradek: wait he changed the whole os?
10:05 AM cradek: O dpm
10:05 AM cradek: oops
10:05 AM skunkworks: No..
10:05 AM cradek: I don't get what he did
10:05 AM skunkworks: that took a bit to figure out
10:06 AM skunkworks: he started with r1 stretch iso - then update/upgraded to 2.7.10 (is what I get)
10:07 AM skunkworks: r1 (2.7.9) install to start with and it appeared to work, been away for holiday and now I did update/upgrade to 2.7.10
10:07 AM cradek: so he might've upgraded lots of stretch packages
10:08 AM cradek: and now there's "some error message about DMI" (or was it always there?)
10:08 AM skunkworks: right - don't know.
10:08 AM cradek: he can sure downgrade linuxcnc from 2.7.10 to 2.7.9 (with apt-get) but of course that won't help with whatever this problem is
10:09 AM skunkworks: I hope to have gotten that point across
10:12 AM skunkworks: bbl
11:21 AM jepler: user can always reinstall from the -r1 image.
11:21 AM jepler: Is there some reason I shouldn't move the git repos for the live image building under the linuxcnc organization?
11:21 AM jepler: https://github.com/jepler/stretch-live-build and https://github.com/jepler/live-wrapper
11:30 AM cradek: I think it would be nice if you moved them
11:35 AM jepler: OK, transferred
01:49 PM jepler: cradek: so on monday I'd like you to first make glo reject all pushes with a message that we're in the midst of the switch to github
01:49 PM jepler: cradek: then after you've done that, I'll add everyone we identified on github's list of people who can push directly to github.com/linuxcnc/linuxcnc
01:50 PM cradek: ok, I can surely figure out which hook to use in just a few tries
01:50 PM jepler: cradek: then I'd like you to switch the push rejection to say that the switch to github is complete
01:50 PM jepler: then at your discretion and timetable we can decommission glo
01:51 PM cradek: sounds good
01:51 PM jepler: cradek: oh I forgot, after disabling push, you'd disable the mirror to github except make sure to do it one last time so everything's there
01:51 PM cradek: right
01:52 PM cradek: the rejection message can even give the shell commands to change origin
01:52 PM jepler: yes
01:55 PM jepler: I think you can use the pre-receive hook for this
01:56 PM jepler: I've updated the #289 checklist and noted who I think can/should do each step
02:19 PM jepler: 2009: we switched to self-hosted git; 2012: I set up github/jepler/linuxcnc-mirror after a "short" glo outage; 2014: I announced I would dedicate time to reviewing pull requests on linux-mirror; 2015: we moved jepler/linuxcnc-mirror to linuxcnc/linuxcnc; 2017: we retire glo and use github as our git hosting
02:20 PM cradek: jeez, two big changes in not even 10 years
02:20 PM jepler: in the 2009 -board list archives, I did present github and gitorious as two alternatives for git hosting
02:22 PM jepler: https://emergent.unpythonic.net/files/sandbox/git-comparison-2009.txt
02:24 PM jepler: (A) github has proven reliable over 8 years (D) we never branded glo (E) github still has no ads over 8 years (F) hm I didn't think about this issue but I never hear of anybody having trouble with github BW limits
02:29 PM jepler: github's standard in 2017 is "significantly excessive in relation to other GitHub customers" https://help.github.com/articles/github-terms-of-service/#7-excessive-bandwidth-use
03:34 PM cradek: jepler: about D) I thought you meant having it under our domain name
03:35 PM jepler: cradek: mmmmaybe
03:52 PM seb_kuzminsky: i think being on github.com will be better for our project than being on linuxcnc.org. more visibility, more casual developers
03:54 PM cradek: yeah I think my feelings about that tradeoff have flipped in those nearly-10-years
04:07 PM jepler: just like being on SF was a boon circa 2005
04:12 PM seb_kuzminsky: “In matters of principle, stand like a rock; in matters of taste, swim with the current.” —Thomas Jefferson.
04:18 PM jepler: I don't get why statements like this one are taken to be logical paradoxes: "The smallest positive integer that cannot be named by a string of at most 90 characters."
04:19 PM jepler: Lots of statements don't name integers (like "this cat"), and lots that appear by their syntax to name integers don't "The integer that is equal to the square root of 15"
04:22 PM seb_kuzminsky: Floor of 1/2 of the smallest positive int that can't be named by a string of <= 90 chars.
04:24 PM jepler: "the smallest positive integer that is too young to legally drink in nevada"
04:24 PM seb_kuzminsky: nevada has laws? i though it was a libertarian mad max utopia
04:25 PM Tom_L: no, that's California
04:26 PM jepler: "There are few restrictions on the sale and consumption of alcohol in Nevada except for age." -- wikipedia page on the Beery Paradox
04:26 PM jepler: er, I mean, Alcohol laws of Nevada
04:26 PM seb_kuzminsky: i think nevada is where i saw the drive-through margarita vending booth in a strip mall
04:26 PM jepler: "State law renders public intoxication legal, and explicitly prohibits any local or state law from making it a public offense on the [Las Vegas, I assume] strip"
04:27 PM jepler: "Open containers are allowed on the Las Vegas Strip and the Fremont Street Experience.[citation needed]"
04:27 PM jepler: (how do you properly cite that there's no law prohibiting something hm)
04:27 PM Tom_L: also in NewOrleans iirc
04:27 PM seb_kuzminsky: "congress shall make no law..."
04:30 PM jepler: https://dylankatz.com/NTLM-Hashes-Microsoft%27s-Ancient-Design-Flaw/ "After researching this further, I realized that this bug was first reported to Microsoft in 1997, making it older than I am."
04:30 PM jepler: aside from how young kids are these days, ...
04:30 PM seb_kuzminsky: 19... 97 you say?
04:30 PM seb_kuzminsky: i hope whatever bug tracker we had in 97 is long gone
04:34 PM Tom_L: wonder where linux was at that point in time
04:37 PM cradek: I was certainly using it every day
04:37 PM seb_kuzminsky: same
04:37 PM Tom_L: what flavors were available then?
04:37 PM Tom_L: redhat?
04:38 PM seb_kuzminsky: slackware for me
04:38 PM seb_kuzminsky: not sure what else was on the table
04:38 PM seb_kuzminsky: yggrasil?
04:38 PM cradek: I don't remember
04:38 PM seb_kuzminsky: *yggdrasil
04:38 PM cradek: yggdrasil was probably pretty dead by then
04:38 PM jepler: In 1997 I was using Fedora
04:38 PM ve7it: http://ve7it.cowlug.org/emc.html notes from 2001
04:38 PM jepler: er not Fedora but Red Hat Linux
04:39 PM ve7it: rh 6.2
04:39 PM jepler: before that I used MCC and SLS, then eventually went to Ubuntu with LinuxCNC, and then helped lead LinuxCNC over to Debian..
04:39 PM seb_kuzminsky: ve7it: that's awesome
04:39 PM seb_kuzminsky: [root@cnc1 src]# ncftp ftp.kernel.org
04:39 PM seb_kuzminsky: ncftp / > get pub/linux/kernel/v2.2/linux-2.2.14.tar.gz
04:40 PM jepler: I remember ncftp! It's apparently still in Debian to this day
04:40 PM seb_kuzminsky: aww, he's got both smp and up kernels in his lilo.conf, that's adorable
04:40 PM jepler: (but I use lftp now)
04:40 PM seb_kuzminsky: # have a cup of coffee, a build takes 10-20 minutes
04:41 PM seb_kuzminsky: brb coffee
04:42 PM ve7it: :} good idea
04:43 PM ve7it: # some stats on system (PP200 dual/64mb and 1.2gb of scsi 2 drive)
04:43 PM seb_kuzminsky: heh
04:44 PM ve7it: those were the days of big chunks of silicon
04:44 PM seb_kuzminsky: these days we need more ram than you had disk, just to run a web browser
04:45 PM Tom_L: shame how resources get wasted
04:45 PM ve7it: I think I still have that motherboard around here somewhere... dual cpu's were cool
04:45 PM Tom_L: you should see dos6.22 run on a modern pc with ssd
04:46 PM jepler: related: the history of most computers I bought from ca. 1992 to 2012: https://emergent.unpythonic.net/01327067482
04:47 PM seb_kuzminsky: ooh, jepler had the AMD 386, was it the 40 MHz version? :-) i had that for a while, but i didn't keep notes
04:47 PM jepler: seb_kuzminsky: yes pretty sure it was the 40 MHz version
04:48 PM seb_kuzminsky: 387 assembly was my first introduction to rpn
04:48 PM cradek: I had a 386/40+387/40 and the 387 made xmountains draw *much* faster
04:49 PM jepler: I'd forgotten all about xmountains
04:49 PM jepler: oh look it's still in debian
04:49 PM jepler: 32.3kB download
04:49 PM jepler: huh I don't get what all the buzz is about :-P
04:49 PM cradek: heh
04:49 PM seb_kuzminsky: that's a blast from the past
04:50 PM cradek: Tom_L: see what you have done
04:51 PM seb_kuzminsky: JT wants none of it
04:51 PM Tom_L: well at least i didn't bring up CPM
04:52 PM ve7it: northstar basic rules
04:52 PM ve7it: with a 2k os in eprom
04:52 PM seb_kuzminsky: if you bring up cpm you have to submit a pr for it at the same time
04:53 PM Tom_L: i'm afraid it was a bit before my time...
04:55 PM jepler: why, just this week I was contributing to a piece of software for the Commodore 64. the author uses github for collaboration (and the software builds on unix, not on the commodore, don't be absurd)
04:56 PM seb_kuzminsky: do you even have a commodore 64?
05:01 PM jepler: nah
05:01 PM jepler: .. but the emulators are good
05:01 PM seb_kuzminsky: heh
05:04 PM jepler: a colleague of mine actually does have lots of working retrocomputers including a VIC-20 and an SX-64
05:04 PM jepler: oh hey I think it's time go to AFK
05:04 PM seb_kuzminsky: seeya jepler
06:05 PM cradek: omg touch-off undo
06:06 PM cradek: that would be tricky to get right, but would sure be nice
06:06 PM * cradek just worked outside for 10 minutes and that was enough
06:07 PM Tom_L: 104 F here
06:08 PM cradek: looks like 102.5/33% here
06:08 PM skunkworks: we got 6+ inches of rain yesterday - more tonight
06:08 PM cradek: wow
06:09 PM cradek: that sounds like a minor disaster
06:09 PM Tom_L: hope you live upstream
06:09 PM cradek: if wunderground is right, we're about to get a little bit of rain and then a break from the heat
06:10 PM skunkworks: lots of mud slides and road closings.. Work had a lot of water in the plant.
06:12 PM seb_kuzminsky: cradek: i imagine separate undo stacks for each (axis, tool, coordinate-system) tuple
06:12 PM skunkworks: you could undo and redo an unlimited amount? :)
06:12 PM seb_kuzminsky: maybe when the touch-off dialog pops up there's a little pull-down menu on the text-entry field that lets you select previous touch-off values for that tuple
06:13 PM skunkworks: the problem is - would you really know which offset was the one you wanted? I wouldn't..
06:13 PM cradek: unfortunately the offsets can be set many ways, so I don't think just remembering previous values of that gui field is going to cut it
06:13 PM seb_kuzminsky: i bet 1 level undo would fix all the accidental-touchoff mistakes i've made, but more would not be much harder
06:14 PM seb_kuzminsky: i think the common mistake i make is touch off one axis where i want it, then position a second axis where i want it, then accidentally touch off the first axis there
06:14 PM seb_kuzminsky: in that situation i'd want to go back to the previous touchoff value on the first axis, and then proceed to touch off the second axis as normal
06:14 PM Tom_L: yeah that's bit me too
06:14 PM cradek: it's tempting to try to do it at the varfile/tooltable level
06:14 PM seb_kuzminsky: hmm
06:15 PM cradek: some touchoffs even change more than one thing
06:15 PM seb_kuzminsky: i was tempted *not* to do it at that level, because i felt uncomfortable with what that interface might look like
06:15 PM seb_kuzminsky: pushing it off to the gui lets me not worry about that part ;-)
06:15 PM cradek: you'd have an nml message for revert tool table, or revert var file, and then you'd do an interp resync
06:15 PM cradek: ... or something?
06:15 PM seb_kuzminsky: you might not want to revert the tooltable/varfile as a unit though
06:16 PM cradek: yeah if I sound comfortable and sure I could get it right, you're misreading me or I'm being unclear
06:16 PM seb_kuzminsky: heh
06:16 PM cradek: :-/
06:16 PM seb_kuzminsky: when does touchoff change multiple things?
06:18 PM cradek: one situation I can think of is when you have the coordinate system rotated and touch off X or Y of a tool
06:18 PM cradek: since it will usually have to change both X and Y of the tool table entry
06:18 PM seb_kuzminsky: oohhh