#linuxcnc-devel Logs

Mar 16 2017

08:22 AM JT-Shop: pcw_home: is the main difference between the 7i84 and the 7i66-24 the voltage and current capabilities aside from the I/O count
09:11 AM pcw_home: Yes mainly output current
09:14 AM JT-Shop: thanks
09:15 AM JT-Shop: is the 7i66-8 in stock?
09:21 AM pcw_home: I think they are both out of stock unfortunately
09:22 AM JT-Shop: ok, thanks
09:23 AM pcw_home: I have around 18 kits I'm building currently and 7I66-8s are among them
09:25 AM pcw_home: other Mesa high current DC output option is 7I69/7I90 and OPTO22
09:26 AM JT-Shop: thanks
09:40 AM mozmck: pcw_home: is the move done?
09:41 AM pcw_home: no, procrastinating to the last moment as usual
09:41 AM mozmck: ah - I know the feeling
09:41 AM JT-Shop: lol
09:42 AM pcw_home: still fixing stuff on the new building
09:42 AM archivist: best procrastinators on the net :)
09:42 AM archivist: Im terrible too
09:42 AM mozmck: I built 128 linear feet of bookshelves for my Dad's books which I got, and it is not nearly enough.
09:43 AM pcw_home: Im going to have to dump a lot of junk I've accumulated (mostly classic computer crap)
09:43 AM mozmck: Trying to move my office from downstairs to upstairs to make room for kids as well, but it is slow.
09:44 AM archivist: mozmck, my books are barcoded and mapped
09:44 AM JT-Shop: I lost four 500' spools of MTW wire moving stuff to the new shop :(
09:44 AM mozmck: archivist: I plan to do something like that - what software do you use?
09:45 AM archivist: mozmck, all homebrew php
09:45 AM mozmck: I'm thinking of using LibraryThing.com
09:45 AM archivist: http://www.collection.archivist.info/shelfview.php?src=artitle&locid=197
09:45 AM mozmck: The main reason is that it will search and enter all the data based on a title or ISBN
09:46 AM archivist: no isbns on old books and manuals
09:46 AM mozmck: Yeah, and I have a lot of those too, but the title search works really well it looks like.
09:47 AM archivist: I did write a fetch to get data but stopped using it
09:48 AM mozmck: Nice thing is that it will export your library in several formats - so once the data is there, I can switch to something else if I want.
09:48 AM archivist: mine lives in mysql
09:48 AM mozmck: I have *no* use for the "social" features of the site - but you can turn a number of them off.
09:49 AM archivist: doesnt librarything have a cost?
09:49 AM mozmck: Yeah, mysql and php is what I was looking for. I don't have time to write something myself.
09:50 AM mozmck: $25 for a lifetime membership if you go over 200 books
09:50 AM mozmck: I'll probably go over 4000
09:50 AM archivist: my qties are on the front page http://www.collection.archivist.info/
09:51 AM mozmck: :-) That's quite a few!
09:51 AM archivist: took a "while" to enter data
09:52 AM archivist: another 550 manuals turned up a month ago
09:53 AM mozmck: I see you have a lot of photographs and papers in there - I don't plan on cataloging any of those here.
09:53 AM archivist: it means I can find and point at examples etc easily
09:53 AM mozmck: Yes, that is nice.
09:54 AM mozmck: Have you seen Zotero?
09:54 AM archivist: it runs an inverted index (like fulltext) or search engines with a few foibles to suit me
09:55 AM archivist: heard of but not looked at
09:57 AM mozmck: It is pretty interesting. I thought about using that for the books.
09:57 AM archivist: mine started out in Paradox on windaz
10:06 AM archivist: and you will find isbn bites as people dont use it properly http://www.collection.archivist.info/searchv13.php?searchstr=BK2755
10:57 AM KGB-linuxcnc: 03Sebastian Kuzminsky 05master 2052cf6 06linuxcnc 10debian/changelog Merge remote-tracking branch 'origin/2.7' * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=2052cf6
11:06 AM jepler: https://lwn.net/SubscriberLink/717221/20a0bb4b4057f86d/ kernelci: For every commit that is made, tests are run with every in-tree defconfig fil e over 260 builds for every commit. The resulting kernels are booted on over 100 different boards
11:06 AM jepler: that must take a heck of a lot of iron
11:12 AM jepler: every time I see the subject line "Suggestions wanted - how to control robot" I imagine the mailing list subscriber has locked themselves in a closet to escape a rampaging robot, and is looking for survival ideas ...
11:25 AM cradek: me too, now
11:26 AM seb_kuzminsky: heh
11:45 AM KGB-linuxcnc: 03Sebastian Kuzminsky 05master c8dd9db 06linuxcnc 10docs/src/gui/gmoccapy.txt docs: fix gmoccapy asciidoc markup * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=c8dd9db
04:47 PM KGB-linuxcnc: 03Sebastian Kuzminsky 05master 8b81a2e 06linuxcnc 10docs/man/man9/kins.9 docs: remove invalid ".R" macro from kins.9 manpage * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=8b81a2e
05:06 PM CaptHindsight: http://linuxcnc.org/ looks down
05:09 PM andypugh: www.linuxcnc.org works
05:13 PM andypugh: Sorry, cached version. Move on, nothing to see here…
05:14 PM CaptHindsight: forums work
05:18 PM seb_kuzminsky: i can ssh in to linuxcnc.org
05:19 PM andypugh: Well, that’s fine then. Anyone who can’t get what they want from the site by ssh won’t be fit to use the software anyway :-)
05:19 PM jepler: it's shared hosting, so there's not a reboot button we can poke
05:19 PM jepler: even from the shared hosting shell you can't curl http://linuxcnc.org/ though
05:20 PM jepler: give it some time to sort itself out, otherwise we'll raise the swp signal and have him file a ticket on dreamhost
05:20 PM seb_kuzminsky: ah, that was you
05:23 PM CaptHindsight: I played around a bit more with the latency tests and the LiveCD installs..
05:24 PM CaptHindsight: I noticed that after a fresh install the latency test has the worst scores the first time you run them and beat on the system
05:25 PM CaptHindsight: scores are always better on the next runs no matter how long you let them run
05:26 PM seb_kuzminsky: try 'echo 1 | sudo tee /proc/sys/vm/dropcaches', then rerun the latency test
05:27 PM seb_kuzminsky: it may be it's better on every run after the first because the code is cached
05:27 PM CaptHindsight: reboots, complete power downs and restarts don't bring the scores back up to what I see on the very first test
05:27 PM seb_kuzminsky: huh, that's weird
05:27 PM CaptHindsight: cache is what i first thought as well
05:28 PM CaptHindsight: I've been running tests for at least a week between restarts
05:29 PM CaptHindsight: I've only installed the system twice so far on the test system
05:30 PM CaptHindsight: I'll see what happens over the next few months
05:31 PM CaptHindsight: it's only a different of maybe 10uS..
05:31 PM CaptHindsight: 20 something vs 30 something uS
05:50 PM jepler: probably the secret CPU microcode sends a packet to the NSA the first time you run linuxcnc, but not after that
05:50 PM jepler: .. I read about it on the internet
05:53 PM seb_kuzminsky: i enjoyed the description of the security measures taken during the zcash parameter generation ceremony: https://soundcloud.com/zcashreview/zcash-review-episode-2-design-of-the-ceremony
05:53 PM seb_kuzminsky: around 3:15
06:27 PM KGB-linuxcnc: 03Dewey Garrett 05dgarr/external_offsets 83d7e72 06linuxcnc 10(63 files in 13 dirs) External Axis Offset hal pins * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=83d7e72
06:31 PM dgarr: jepler: thanks for fixing the race in userspace-count-names
06:31 PM dgarr: jepler: btw, i was able to reproduce the axis gui memleak on debian stretch (in a vm) described: https://forum.linuxcnc.org/21-axis/32469-axis-memory-leak-2-8-0-pre1#89478
06:36 PM seb_kuzminsky: dgarr: do you know if the leak is in axis or in one of the libraries?
06:39 PM dgarr: no clue
06:44 PM seb_kuzminsky: 99 problems
06:47 PM dgarr: my test was sim, plain kernel, fresh install 4.9.0-1 amd64 #1 SMP Debian 4.9.4-2 (2012-01-12)
06:49 PM seb_kuzminsky: and did you just watch axis' memory usage with top or something?
06:52 PM dgarr: top and also grep VmSize /proc/the_python_process_id/status
06:56 PM linuxcnc-build: build #2421 of 4017.deb-wheezy-amd64 is complete: Failure [4failed shell_2] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/4017.deb-wheezy-amd64/builds/2421 blamelist: Sebastian Kuzminsky <seb@highlab.com>
06:56 PM seb_kuzminsky: sigh
06:56 PM seb_kuzminsky: W: Failed to fetch http://linuxcnc.org/dists/wheezy/Release.gpg Could not connect to linuxcnc.org:80 (69.163.224.31). - connect (110: Connection timed out)
07:04 PM linuxcnc-build: build #1676 of 4014.deb-wheezy-rtpreempt-i386 is complete: Failure [4failed shell_2] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/4014.deb-wheezy-rtpreempt-i386/builds/1676 blamelist: Sebastian Kuzminsky <seb@highlab.com>
07:41 PM linuxcnc-build: build #1092 of 4021.deb-jessie-i386 is complete: Failure [4failed shell_2] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/4021.deb-jessie-i386/builds/1092 blamelist: Sebastian Kuzminsky <seb@highlab.com>
08:15 PM linuxcnc-build: build #1711 of 4015.deb-wheezy-rtpreempt-amd64 is complete: Failure [4failed shell_2] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/4015.deb-wheezy-rtpreempt-amd64/builds/1711 blamelist: Sebastian Kuzminsky <seb@highlab.com>
08:30 PM jepler: seb_kuzminsky: fwiw the front page of linuxcnc.org just loaded for me
08:30 PM jepler: dgarr: thanks for the info
08:30 PM jepler: time was I could leave AXIS running for 2 weeks and memory usage was stable. groan.
08:33 PM jepler: is "stretch" a key factor here?
08:34 PM jepler: "virt" usage is stable for minutes (unchanged by even the last digit) with tip of master on jessie
08:35 PM jepler: running axis.ini, not the config attached to the forum post, but I see dgarr said he used sim
08:35 PM dgarr: i think has only been noted on stretch, op reports in thread he installed jessie and then had no problems
08:37 PM dgarr: i tested on 64bit stretch vm with axis.ini and saw mem growth just at idle with std program (not running)
08:38 PM jepler: no need to do any actions on the axis window?
08:39 PM dgarr: not that i recall , i think i just did estop-off, machine-on
08:41 PM jepler: hm
08:42 PM jepler: in a squeeze schroot, virt is growing slowly, I've seen it go from 528052 to 536728. it seems to only grow when I change desktops
08:58 PM jepler: is that different or the same as what you're seeing, dgarr?
08:59 PM dgarr: i now see 480340 to 519068 in 14 minutes
09:00 PM dgarr: i have not changed desktops with the vm
09:00 PM dgarr: s/with/within/
09:01 PM jepler: OK, then I don't think we see the same thing yet
09:01 PM jepler: updating packages in my schroot
09:09 PM jepler: still don't see anything when not interacting with the window
09:09 PM jepler: when I rotate or pan I do see memory growth, again in stretch only and not in jessie
09:57 PM jepler: I eventually got it in a state where memory grows pretty steadily. showing the text-only dro triggers it for me.
09:59 PM jepler: valgrind indicates the leak is occurring inside the text widget 'delete' command, and indeed a pretty simple standalone python program is triggering a big memory leak for me, stretch only. https://emergent.unpythonic.net/files/sandbox/tkileak.py
10:02 PM jepler: and directly in tcl https://emergent.unpythonic.net/files/sandbox/tclleak.tcl
10:04 PM jepler: again stretch only
10:44 PM jepler: dgarr: if you can please test with https://emergent.unpythonic.net/files/sandbox/0001-axis-avoid-unbounded-memory-growth-in-text-widgets-o.patch
10:44 PM jepler: I know I've found *something* but I'm not 100% sure if it's what you and the forum user encountered
10:46 PM jepler: 'night
11:16 PM dgarr: jepler: i tested your patch (at master 8b81ade4b current) and it stopped leak, zero VmSize increase in a test identical to my earlier test (stretch,vm,64,axis.ini) -- superlative work, thanks
11:16 PM dgarr: (25minute test)
11:29 PM KGB-linuxcnc: 03Jeff Epler 052.7 0428111 06linuxcnc 10share/axis/tcl/axis.tcl axis: avoid unbounded memory growth in text widgets on stretch * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=0428111
11:29 PM KGB-linuxcnc: 03Jeff Epler 05master bdea870 06linuxcnc 10share/axis/tcl/axis.tcl Merge remote-tracking branch 'origin/2.7' * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=bdea870
11:29 PM jepler: dgarr: thanks for bringing it up and for the testing. goodnight.
11:29 PM jepler: (that was fun, why do I have a splitting headache?)