#linuxcnc-devel | Logs for 2014-07-14

[08:33:17] <cradek> seb_kuzminsky: a test! that's awesome.
[08:33:31] <cradek> and you even read the docs and finished the implementation.
[08:36:50] <jepler> huh, TIL you can remove a package with 'apt-get install', e.g., sudo apt-get install scons-
[08:38:05] <cradek> I bet that's handy in twisty dependency situations
[08:38:31] <jepler> yeah, you can mix it with install requests and even with (dist-)upgrade
[08:38:41] <jepler> apt-get dist-upgrade newpackage oldpackage-
[09:01:57] <skunkworks> logger[psha],
[09:01:58] <logger[psha]> skunkworks: Log stored at http://psha.org.ru/irc/%23linuxcnc-devel/2014-07-14.html
[09:16:05] <KGB-linuxcnc> 03Jeff Epler 05master ce3872f 06linuxcnc 10src/hal/drivers/mesa-hostmot2/bspi.c 10src/hal/drivers/mesa-hostmot2/sserial.c 10src/hal/drivers/mesa-hostmot2/uart.c hostmot2: fix 'set but unused' compiler diagnostics * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=ce3872f
[09:16:05] <KGB-linuxcnc> 03Jeff Epler 05master 2411173 06linuxcnc 10src/hal/drivers/mesa-hostmot2/bspi.c 10src/hal/drivers/mesa-hostmot2/hostmot2.h 10src/hal/drivers/mesa_7i65.comp hostmot2: bspi: Use proper function pointers * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=2411173
[09:16:05] <KGB-linuxcnc> 03Jeff Epler 05master ca63d6a 06linuxcnc 10src/hal/drivers/mesa-hostmot2/bspi.c hostmot2: bspi: fix 'set but unused' compiler diagnostics * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=ca63d6a
[09:16:09] <KGB-linuxcnc> 03Jeff Epler 05master d2e8854 06linuxcnc 10src/hal/drivers/mesa-hostmot2/tram.c hostmot2: fix 'set but unused' compiler diagnostics * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=d2e8854
[09:17:26] <KGB-linuxcnc> 03Jeff Epler 05jepler/rtos-uspace-apis a773b76 06linuxcnc 27 commits pushed, 101 file changed, 032(+)
[09:17:32] <jepler> linuxcnc builds with gcc 4.9 over here (debian testing recently transitioned to 4.9.0 as the default compiler)
[09:27:14] <seb_kuzminsky> it's funny that it was just hostmot2 that had trouble with gcc 4.9
[09:27:21] <seb_kuzminsky> did i say funny? i meant sad
[09:28:53] <seb_kuzminsky> cradek: we have this section in our docs called http://www.linuxcnc.org/docs/2.6/html/gcode/overview.html#_numbered_parameters_a_id_sub_numbered_parameters_a, which confused me when writing the g43{,.2} test
[09:30:22] <seb_kuzminsky> #5401-#5409 are called "Tool Offsets", but they hold the TLOs for the most recently M6'd tool, no matter if any G43s have been run
[09:30:56] <seb_kuzminsky> the only way i could find to see what the G43s did is to check #5420-#5428 "Current Position"
[09:31:36] <jepler> seb_kuzminsky: 4.8 also had the same diagnostics
[09:33:02] <seb_kuzminsky> cradek: i think the code is probably fine, i think my problem was that the docs for the TLO params are a bit terse
[09:33:10] <seb_kuzminsky> jepler: thanks for fixing it
[09:36:17] <jepler> seb_kuzminsky: I hope I got it all correct
[09:36:43] <jepler> ugh I shouldn't have gone looking in libnml
[09:38:30] <KGB-linuxcnc> 03Jeff Epler 05master f4f2104 06linuxcnc 10src/libnml/buffer/tcpmem.cc 10src/libnml/cms/tcp_srv.cc libnml: Disable "strict aliasing" * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=f4f2104
[09:39:51] <jepler> there, I'll scoop it all under the rug
[09:42:39] <seb_kuzminsky> heh
[09:43:04] <jepler> the real worry is not that strict-aliasing BS, but the use of u_long everywhere for values in "portable" structures
[09:43:13] <jepler> .. since u_long is 64 bits on amd64 platforms
[09:43:51] <cradek> I think remote nml is confirmed broken between machines with different sizes
[09:44:01] <jepler> yeah
[09:44:11] <jepler> someone who cares should fix it
[09:44:19] <jepler> since single-machine mode works just excellent on amd64
[09:44:22] <cradek> some people on the mailing list futzed with it but didn't manage to fix it (possibly being distracted by the warnings and not understanding what the real problems were)
[09:44:53] <jepler> sigh
[09:54:32] <cradek> hey, I even tested g43.2 while rotated, yay me
[10:07:19] <cradek> ERROR: interp param #5420 = -1.572605, expected -1.572605
[10:07:24] <cradek> (sigh)
[10:18:25] <seb_kuzminsky> cradek: is that in my new test? did i botch the "floats are close enough to equal" test again?
[10:19:04] <KGB-linuxcnc> 03Chris Radek 05cradek/multi-tlo 11ca375 06linuxcnc 10tests/tlo/test.sh Start with a clean var file each time, since we might have rotated * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=11ca375
[10:19:04] <KGB-linuxcnc> 03Chris Radek 05cradek/multi-tlo c5b5725 06linuxcnc 10tests/tlo/test-ui.py The interpreter doesn't report very many decimal places to the UI * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=c5b5725
[10:19:04] <KGB-linuxcnc> 03Chris Radek 05cradek/multi-tlo 479fdfa 06linuxcnc 10tests/tlo/test-ui.py Also test clear/set TLO and summing TLO while rotated * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=479fdfa
[10:20:50] <cradek> seb_kuzminsky: you tested nice round numbers, so it didn't matter - I was testing a calculated value against what the interpreter reports, which is quite chopped off
[10:21:14] <seb_kuzminsky> i see the rotation, that's a nice addition to the test
[10:21:46] <seb_kuzminsky> have i linked to the "secret robot internet" comic that makes fun of floats in the past week or so?
[10:21:53] <cradek> haha yes
[11:12:02] <CaptHindsight> was there any RTAI kernel to test on wheezy besides the linux-image-3.4.55-rtai-2 with rtai-modules-3.4.55-rtai-2 ?
[11:12:20] <CaptHindsight> from deb http://linuxcnc.org precise base
[11:13:27] <cradek> CaptHindsight: yes, http://highlab.com/~seb/linuxcnc/rtai-debs/dists/wheezy/main/binary-i386/
[11:14:00] <cradek> there's also a wheezy live/install image with this kernel and rtai: http://www.linuxcnc.org/binary.hybrid.iso
[11:14:39] <cradek> in particular we're most interested in testing linux-image-3.4-9-rtai-686-pae_3.4.55-3linuxcnc_i386
[11:16:31] <CaptHindsight> I'll give those a try later today
[11:16:55] <cradek> I think the usb on skunkworks's machine still doesn't work with that kernel
[11:20:35] <CaptHindsight> BIOS and video drivers really make a big difference
[11:20:52] <cradek> well the stock debian kernels work fine
[11:21:33] <CaptHindsight> oh for USB, I meant for latency results on the boards here
[11:21:47] <cradek> oh definitely
[11:23:02] <CaptHindsight> one BIOS on a board here still in production but pre-EFI doesn't recognize some of the install images
[11:23:33] <CaptHindsight> I have 6 distros on a 3TB drive I now swap between systems
[11:24:49] <CaptHindsight> and what installs great with no hiccups on a really new mainboard might have really poor latency, yet run great with <8K on a older board
[11:29:34] <CaptHindsight> when memleak tweaks a RTAI kernel and the video drivers we get <5K on most boards, the Livecd or 3.4.55 RTAI on Mint Debian Mate (the Gnome2 fork) will be 8-13K
[11:30:34] <CaptHindsight> there's something about Ubuntu or Mint with gnome3 that slows it down further
[11:30:59] <CaptHindsight> I have to compare the video driver source they used
[11:46:45] <cradek> I don't really care about 5 vs 8 vs 13k, I care about a kernel that runs adequately (< 25k with working usb/ps2/video/network/pata/sata/scsi) on the widest variety of hardware possible
[12:20:32] <CaptHindsight> I understand and agree, just reporting recent observations
[12:21:39] <cradek> cool, sorry, I didn't mean to sound so disagreeable.
[12:23:41] <Tom_itx> excuse: it's Monday
[12:30:02] <CaptHindsight> I have to decide on what new cpu, maniboard, kernel and distro to use for a guberment Linuxcnc application
[12:53:16] <jepler> surely somebody's got a cache of new old stock commodore 128Ds with a built in double-sided floppy and CP/M mode
[12:53:42] <seb_kuzminsky> maybe peter can whip one up on a 7i34-200
[12:53:47] <jepler> oh and don't forget 80 column text, if you can field a RGBI monitor.
[12:54:24] <seb_kuzminsky> "imagine a beowulf cluster of those!"
[12:55:00] <jepler> 100% compatible with CP/M 3.0 applications such as Turbo Pascal and WordStar
[12:57:24] <jepler> I had totally forgotten that the 128 added a numeric keypad and 3 sets of 4 keys above the keyboard. it looks superficially a lot more like an AT keyboard than the Commodore 64 did.
[13:05:37] <CaptHindsight> does NIST still use linuxcnc in any way or their own original branch?
[13:37:48] <cradek> CaptHindsight: we don't know
[14:20:57] <skunkworks> I wonder if dad still has our old 128's
[14:53:28] <KGB-linuxcnc> 03Francis Tisserant 05v2.5_branch 0397cb1 06linuxcnc 10src/po/fr.po French translation update * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=0397cb1
[15:04:49] <cradek> seb_kuzminsky: how do you feel about multi-tlo in 2.6?
[15:07:01] <seb_kuzminsky> it seems well compartmentalized and straight-forward
[15:07:18] <seb_kuzminsky> i generally dislike new features (except new drivers, they're ok) but maybe just this once
[15:07:24] <cradek> yay
[15:10:35] <seb_kuzminsky> i'm diffing the configs for the debian 3.2 kernel with working skunkusb and the 3.4.55-3linuxcnc rtai kernel where it's broken
[15:11:09] <seb_kuzminsky> 3.4.55-3linuxcnc lacks PCI_IOAPIC support and some ACPI things, that's the first suspicious think i've seen (but i'm only 20% through the 2k line diff
[15:11:35] <seb_kuzminsky> oh, also MSI is turned off for some reason, that seems like a bug
[15:12:25] <skunkworks> seb_kuzminsky, I have not tried the debian kernel yet.. I tried to get it last week and it would not download
[15:12:35] <skunkworks> *stock
[15:13:45] <KGB-linuxcnc> 03Chris Radek 052.6 ea577b1 06linuxcnc 10src/emc/rs274ngc/interp_array.cc 10src/emc/rs274ngc/interp_check.cc 10src/emc/rs274ngc/interp_convert.cc 10src/emc/rs274ngc/interp_internal.hh Allow use of several simultaneous tool offsets with G43.2 H * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=ea577b1
[15:13:46] <KGB-linuxcnc> 03Chris Radek 052.6 b7914b4 06linuxcnc 10src/emc/usr_intf/touchy/mdi.py Touchy support for mdi G43.2 * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=b7914b4
[15:13:46] <KGB-linuxcnc> 03Chris Radek 052.6 1b76410 06linuxcnc 10docs/html/gcode.html G43.1 works with all axes, not just XZ now * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=1b76410
[15:13:48] <KGB-linuxcnc> 03Chris Radek 052.6 454e5b2 06linuxcnc 10docs/html/gcode.html 10docs/src/gcode/gcode.txt Documentation for G43.2 * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=454e5b2
[15:13:52] <KGB-linuxcnc> 03andy pugh 052.6 8f53d01 06linuxcnc 10(6 files) A demo config combining cradek's compounf offset (G43.2) and mah's * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=8f53d01
[15:13:55] <KGB-linuxcnc> 03Sebastian Kuzminsky 052.6 70c9c02 06linuxcnc 10src/emc/rs274ngc/interp_convert.cc g43.2: verify H-word is present, as per the docs * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=70c9c02
[15:13:59] <KGB-linuxcnc> 03Sebastian Kuzminsky 052.6 9b16648 06linuxcnc 10(8 files) tests: add a test of G43 & G43.2 * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=9b16648
[15:14:03] <KGB-linuxcnc> 03Chris Radek 052.6 a1c1853 06linuxcnc 10tests/tlo/test.sh Start with a clean var file each time, since we might have rotated * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=a1c1853
[15:14:07] <KGB-linuxcnc> 03Chris Radek 052.6 97f871c 06linuxcnc 10tests/tlo/test-ui.py The interpreter doesn't report very many decimal places to the UI * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=97f871c
[15:14:11] <KGB-linuxcnc> 03Chris Radek 052.6 a0eaaf1 06linuxcnc 10tests/tlo/test-ui.py Also test clear/set TLO and summing TLO while rotated * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=a0eaaf1
[15:14:15] <KGB-linuxcnc> 03Chris Radek 05master 6ee690c 06linuxcnc Merge branch '2.6' * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=6ee690c
[15:14:34] <seb_kuzminsky> skunkworks: i have this from you:
[15:14:35] <seb_kuzminsky> <skunkworks> cradek, jepler, usb ports don't work with rtai kernel - do work with the debian -rt kernel...
[15:14:49] <seb_kuzminsky> cradek: yay! :-)
[15:15:02] <seb_kuzminsky> skunkworks: did i misunderstand something?
[15:16:04] <skunkworks> yes - that is correct - but I think you wanted me to try the stock non rt debian kernel
[15:16:30] <skunkworks> or isn't that needed as it works with the -rt?
[15:16:35] <seb_kuzminsky> ah, i see
[15:16:48] <seb_kuzminsky> yes, it would be helpful to know if the non-rt kernel has working usb on that harwdare
[15:17:42] <seb_kuzminsky> bbl
[15:21:00] <KGB-linuxcnc> 03Francis Tisserant 05v2.5_branch f4af721 06linuxcnc 10docs/src/hal/basic_hal_fr.txt French doc update: fix float value with old data * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=f4af721
[15:23:34] <KGB-linuxcnc> 03bdebruijn 05cradek/pid-in a4235a6 06linuxcnc 10src/hal/components/pid.c components : change IO pins of pid to IN * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=a4235a6
[15:23:34] <KGB-linuxcnc> 03bdebruijn 05cradek/pid-in 80c46b9 06linuxcnc 10docs/man/man9/pid.9 docs : update manpage pid.9 for use of IN pins * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=80c46b9
[15:24:07] <cradek> seb_kuzminsky: bugfix (imo) picked from machinekit ^^
[15:24:43] <KGB-linuxcnc> 05cradek/multi-tlo 479fdfa 06linuxcnc 04. branch deleted * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=479fdfa
[15:56:14] <KGB-linuxcnc> 03Chris Radek 052.6 80c46b9 06linuxcnc fast forward * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=80c46b9
[15:56:41] <KGB-linuxcnc> 05cradek/pid-in 80c46b9 06linuxcnc 04. branch deleted * 14http://git.linuxcnc.org/?p=linuxcnc.git;a=commitdiff;h=80c46b9
[15:57:54] <seb_kuzminsky> thx cradek & bas de bruijn