Back
[00:04:20] <JT-Shop> ouch
[00:04:38] <JT-Shop> Andy thanks for sorting out that it is a bug
[00:15:31] <seb_kuzminsky> yikes, i use halui for jogging on my machine! i hope andypugh fixes it before i upgrade!
[00:16:25] <andypugh> Can't see the problem so far...
[00:17:47] <seb_kuzminsky> if you write a test that reproduces the problem, you can use git bisect to discover the commit that introduced the bug
[00:18:23] <seb_kuzminsky> plus, we'll have a new test that verifies halui jogging, which would be great!
[00:20:08] <andypugh> Currently my test is using my matrix keyboard driver to control the physical jog buttons. Which is a bit of a problem with the bisect idea...
[00:20:48] <seb_kuzminsky> yeah, a test script is really the way to go, for a variety of reasons
[00:21:04] -!- Dupe has quit [Read error: Connection reset by peer]
[00:21:35] <andypugh> How do you script something like this?
[00:26:12] <seb_kuzminsky> look at (for example) the linuxcncrsh test for the basic setup
[00:27:00] <seb_kuzminsky> but instead of sending mdi commands run "halcmd sets" to simulate button presses
[00:28:05] <seb_kuzminsky> and use "halcmd show" to inspect the axis motor-commands to make sure the expected axis (and only the expected axis!) moves
[00:29:45] <andypugh> Job 1) Sleep. Job 2) Learn shell scripting...
[00:33:40] -!- Valen has quit [Read error: No route to host]
[00:34:42] -!- adb has quit [Read error: Operation timed out]
[00:39:44] <andypugh> If anyone wants to check this, you can see it in the 2.6version in sim/axis
[00:40:47] <andypugh> setp halui.joint.1.select 1 / setp halui.jog.selected.plus 1 / setp halui.jog.selected.plus 0 will jog joint 1
[00:41:12] <andypugh> setp halui.joint.2.select 1 / setp halui.jog.selected.plus 1 / setp halui.jog.selected.plus 0 will _also_ jog joint 1.
[00:42:06] <andypugh> But the joint.2.is-selected pin shows the correct status.
[00:44:30] -!- rob_h has quit [Ping timeout: 264 seconds]
[00:45:58] -!- fragalot has quit [Ping timeout: 256 seconds]
[00:47:06] fragalot is now known as Guest64639
[00:53:52] <andypugh> is there a way to compare files from two different branches?
[00:58:33] <jepler> andypugh: git diff rev1 rev2 -- filename
[00:58:50] <jepler> e.g., git diff v2.5_branch master -- src/Makefile
[01:03:06] <andypugh> Well the only difference in the obvious place to look (halui.cc) is lots of changes of EMC_NMLFILE to emc_nmlfile
[01:08:14] <seb_kuzminsky> that's all i see in halui too
[01:08:21] <seb_kuzminsky> git bisect, man!
[01:08:52] <seb_kuzminsky> you mark one commit as good (origin/v2.5_branch), mark one commit as bad (origin/master)
[01:09:07] <seb_kuzminsky> then show git how to tell if another commit is good or bad (by compiling and running the test script)
[01:09:12] <seb_kuzminsky> then you start it, and go to sleep
[01:09:20] <seb_kuzminsky> in the morning, git will tell you which commit introduced the bug
[01:09:25] <andypugh> I have done it in the past. I am prepared to do it again. But not at 0100 on a work night. I have a meeting in 7.5 hours.
[01:09:30] <seb_kuzminsky> heh
[01:09:33] <seb_kuzminsky> i know that feel
[01:10:10] <andypugh> I am happy enough to do it manually, now that I know how simple it is to spot the problem.
[01:10:33] <seb_kuzminsky> but will you do it manually on every commit from now on? ;-)
[01:11:33] <andypugh> No, because I am lacksidaisical and not a proper programmer.
[01:12:41] <seb_kuzminsky> you're a fine programmer - you've written many deep & useful programs for linuxcnc
[01:13:08] <seb_kuzminsky> the smart serial stuff is serious business
[01:15:42] -!- asdfasd has quit [Ping timeout: 252 seconds]
[01:16:17] <andypugh> I now have a keyboard device connected by smart-serial :-). Links into system just like any other keyboard.
[01:16:54] <seb_kuzminsky> "just like any other keyboard" - meaning through the linux Input system?
[01:17:09] <andypugh> (In fact, yesterday I had three, called tom, dick and harry, all reading the same scancodes from the matrix, but typing different letters into the console.)
[01:17:17] <seb_kuzminsky> neat!
[01:19:18] <andypugh> http://pastebin.com/BVzCD33E
[01:19:46] <seb_kuzminsky> best of both worlds
[01:20:07] <seb_kuzminsky> see - you can write code like that, you can write tests for it ;-)
[01:20:41] <andypugh> And if I had started when I first said it was too late to start, I would be well on the way..
[01:20:47] <seb_kuzminsky> haha
[01:21:28] <andypugh> I will make a start tomorrow if nobody else finds out where the problem is.
[01:21:48] <andypugh> (It would be nice if someone else could check that they see the issue too)
[01:21:53] <seb_kuzminsky> i
[01:21:55] <seb_kuzminsky> o
[01:21:57] <seb_kuzminsky> uh
[01:22:03] <seb_kuzminsky> i'll check and report here
[01:22:16] <andypugh> Right, night all
[01:22:27] -!- andypugh has quit [Quit: andypugh]
[01:22:52] * seb_kuzminsky is a spaz
[01:23:21] -!- tjb1 has quit [Ping timeout: 245 seconds]
[01:28:00] -!- FinboySlick has quit [Quit: Leaving.]
[01:29:09] -!- hdokes has quit [Quit: On the other hand, you have different fingers.]
[01:35:09] -!- yhoje has quit [Quit: Leaving.]
[01:36:35] -!- jfire has quit [Quit: Leaving.]
[01:51:24] -!- yhoje has quit [Quit: Leaving.]
[01:56:27] -!- tjb1 has quit [Ping timeout: 260 seconds]
[02:09:03] -!- sumpfralle has quit [Ping timeout: 260 seconds]
[02:12:26] -!- yhoje has quit [Ping timeout: 246 seconds]
[02:23:48] -!- yhoje has quit [Ping timeout: 250 seconds]
[02:23:50] -!- FinboySlick has quit [Quit: Leaving.]
[02:27:10] <cradek> interestingly, I see no meaningful code differences in halui between v2.5_branch and master
[02:27:13] -!- yhoje1 has quit [Read error: Connection reset by peer]
[02:28:05] <cradek> I was going to snarkily say that bisecting is unnecessary when you know what file the problem is in and there's only been once change, but ... after looking, I see that change didn't cause it!
[02:29:01] <seb_kuzminsky> yeah, isnt that fun
[02:29:06] <cradek> fwiw, I recommend turning on TASK_ISSUE debug to quickly see whether it's motion or ui
[02:29:13] <seb_kuzminsky> halui links a bunch of stuff
[02:29:58] <cradek> the selected-axis stuff smells bad. it uses the same int both as flag and joint number I think
[02:30:37] <cradek> (I haven't even tried to reproduce it)
[02:32:16] <seb_kuzminsky> only andy has repro'd it so far
[02:32:40] <cradek> I had the hubris to think I could go right to the code and fix it
[02:33:34] <seb_kuzminsky> it's hackspace night! we're gonna try 3d printing motor mounts for the cnc conversion of our little benchtop mill (g0407
[02:33:49] <seb_kuzminsky> 0704? something like that
[02:33:51] <cradek> plastic motor mounts!?
[02:33:55] <seb_kuzminsky> hahahah
[02:34:18] <seb_kuzminsky> i hope they last long enough to machine replacements out of Al
[02:34:32] <cradek> if only you had access to another mill!
[02:34:48] <seb_kuzminsky> heh
[02:44:03] <skunkworks> that isn't the point ;)
[02:44:33] <seb_kuzminsky> the prints that our guy jim turpin produces are surprisingly strong
[02:44:45] <cradek> cool, I'd like to see that
[02:44:53] <seb_kuzminsky> we just dropped a 30-pound weight on one and it'd didn't fail...
[02:45:34] <seb_kuzminsky> i just repro'd the bug
[02:45:49] <seb_kuzminsky> selecting joint 2 and then jogging the selected axis jogs joint 1
[02:50:08] <seb_kuzminsky> that's with sim/axis/axis.ini
[02:51:52] <cradek> is the nml command right or wrong?
[02:52:01] <cradek> it's just gotta be wrong
[02:52:41] <seb_kuzminsky> it's right when axis ui jogs it with pgup/pgdn
[02:53:49] <seb_kuzminsky> i
[02:54:15] <seb_kuzminsky> i'm about as interested in writing the test in python now as i am in actually fixing the bug
[02:54:30] -!- yhoje has quit [Read error: Connection reset by peer]
[02:54:32] <cradek> heh
[02:54:38] <cradek> seems hard to fix
[02:55:23] <cradek> do you consider the likeliness of regression when you decide whether to write a test, or do you think it doesn't matter, or that you just can't know it?
[02:55:38] <cradek> er, I meant it seems hard to test
[02:55:41] -!- tmcw has quit [Remote host closed the connection]
[02:55:43] <seb_kuzminsky> i really want 100% test coverage
[02:56:14] <seb_kuzminsky> partially just to know, and partially to allow refactoring with confidense
[02:56:39] <seb_kuzminsky> seems easy to test, imo
[02:57:02] <seb_kuzminsky> i'm trying to do it with the python interface instead of the linuxcncrsh interface, which is making it harder ;-)
[02:57:06] <cradek> can you get ahold of the issued nml message? or will you watch the motion?
[02:57:12] <seb_kuzminsky> i'll watch motion
[02:57:15] <cradek> wow
[03:00:03] -!- RangerRick has quit [Remote host closed the connection]
[03:00:36] -!- tmcw has quit [Ping timeout: 248 seconds]
[03:02:30] <seb_kuzminsky> hm, the linuxcnc python module does not provide an interface to hal
[03:02:53] <seb_kuzminsky> ah, but the "hal" python module does
[03:07:24] -!- markvandenborre has quit [Ping timeout: 264 seconds]
[03:07:39] -!- Keknom has quit [Quit: Leaving.]
[03:07:47] <seb_kuzminsky> cool, that works
[03:08:54] -!- gimps has quit [Read error: Connection reset by peer]
[03:13:51] -!- zzolo has quit [Quit: zzolo]
[03:17:54] -!- Valen has quit [Ping timeout: 245 seconds]
[03:29:27] -!- Valen has quit [Quit: Leaving.]
[03:31:36] -!- FinboySlick has quit [Quit: Leaving.]
[03:53:54] -!- yhoje has quit [Ping timeout: 258 seconds]
[03:56:35] -!- yhoje has quit [Client Quit]
[04:00:35] -!- yhoje1 has quit [Ping timeout: 260 seconds]
[04:02:19] <KGB-linuxcnc> 03matt 05v2.5_branch e7815ae 06linuxcnc 10configs/smithy/ 101240combined_4axis.ini 101315.hal 03naiky_4axis.hal * Add naiky_axis.hal.
[04:02:19] <KGB-linuxcnc> 03matt 05v2.5_branch bc20a39 06linuxcnc * Merge branch 'v2.5_branch' of ssh://git.linuxcnc.org/git/emc2 into v2.5_branch
[04:04:49] -!- tjb1 has quit [Quit: tjb1]
[04:06:09] -!- tmcw has quit [Ping timeout: 258 seconds]
[04:09:42] ink_ is now known as ink
[04:13:29] -!- yhoje has quit [Read error: Connection reset by peer]
[04:29:11] -!- sumpfralle has quit [Ping timeout: 245 seconds]
[04:47:21] <linuxcnc-build> build #28 of precise-i386-realtime-rip is complete: Failure [4failed get-dmesg] Build details are at
http://buildbot.linuxcnc.org/buildbot/builders/precise-i386-realtime-rip/builds/28 blamelist: Matt Shaver <matt@mattshaver.com>
[04:48:48] <seb_kuzminsky> awesome, the precise-rtai builder segfaulted in dmesg
[04:48:53] <seb_kuzminsky> imma reboot it
[04:49:01] <seb_kuzminsky> http://buildbot.linuxcnc.org/buildbot/builders/precise-i386-realtime-rip/builds/28/steps/get-dmesg/logs/stdio
[04:49:35] -!- sumpfralle has quit [Ping timeout: 260 seconds]
[04:52:35] -!- hdokes has quit [Read error: No route to host]
[04:56:29] <seb_kuzminsky> i ran dmesg a couple of times, and each time it segfaulted like that
[04:56:38] <seb_kuzminsky> a reboot fixed it
[04:56:49] * seb_kuzminsky feels uneasy
[04:59:20] <KGB-linuxcnc> 03seb 05master-halui-jog-bug 17e1a22 06linuxcnc 10tests/ 10(8 files) * add a a halui jogging test
[04:59:36] <seb_kuzminsky> that's a test that demonstrates the halui jogging bug in master
[04:59:48] <seb_kuzminsky> it uses the linuxcnc python module and the hal python module
[04:59:58] <seb_kuzminsky> way, way cooler than the linuxcncrsh crap i was using for testing
[05:01:35] <linuxcnc-build> build #825 of checkin is complete: Failure [4failed] Build details are at
http://buildbot.linuxcnc.org/buildbot/builders/checkin/builds/825 blamelist: Matt Shaver <matt@mattshaver.com>
[05:01:56] <seb_kuzminsky> same failure as the precise-rtai-x86 one
[05:06:11] -!- ve7it has quit [Remote host closed the connection]
[05:06:43] -!- tmcw has quit [Ping timeout: 258 seconds]
[05:07:02] <seb_kuzminsky> wow, this halui bug has been around for a while...
[05:07:12] <seb_kuzminsky> i guess not a lot of people run the master branch
[05:07:46] -!- kiplitton has quit [Quit: leaving]
[05:08:09] <seb_kuzminsky> hmm, bisecting past the rebranding is annoying
[05:14:43] <seb_kuzminsky> heh, and bisecting past the introduction of the linuxcnc python module is impossible
[05:15:48] <seb_kuzminsky> heh, and grepping for "python" in the git log is annoying because jepler@unpythonic.net
[05:32:13] -!- joe9_ has quit [Quit: leaving]
[05:43:25] -!- kwallace [kwallace!~kwallace@smb-4.sonnet.com] has parted #linuxcnc-devel
[05:50:48] -!- tjb1 has quit [Ping timeout: 258 seconds]
[05:50:48] tjb1_ is now known as tjb1
[05:50:51] Hewball is now known as hewball
[05:55:23] -!- AR_ has quit [Ping timeout: 260 seconds]
[05:57:19] -!- tjb1 has quit [Ping timeout: 258 seconds]
[06:01:33] -!- Fox_Muldr has quit [Ping timeout: 252 seconds]
[06:09:06] -!- tmcw has quit [Ping timeout: 264 seconds]
[06:11:51] -!- jfire has quit [Quit: Leaving.]
[06:11:55] -!- psha[work] [psha[work]!~psha@195.135.238.205] has joined #linuxcnc-devel
[06:25:16] -!- vladimirek [vladimirek!~vladimire@95.105.250.72] has joined #linuxcnc-devel
[06:25:23] -!- tripod| has quit [Ping timeout: 252 seconds]
[06:33:21] -!- tjb1 has quit [Ping timeout: 245 seconds]
[06:41:01] -!- Tom_itx has quit [Ping timeout: 240 seconds]
[06:50:27] -!- pfred1 has quit [Quit: nite]
[06:58:54] -!- tripod| has quit [Ping timeout: 264 seconds]
[07:02:26] -!- jpk has quit [Ping timeout: 256 seconds]
[07:09:46] -!- tmcw has quit [Ping timeout: 258 seconds]
[07:18:25] -!- R2E4_ has quit [Read error: Connection reset by peer]
[07:20:43] -!- tripod| has quit [Read error: Connection reset by peer]
[07:24:20] -!- hashfail has quit []
[07:26:46] <seb_kuzminsky> strange, the halui-jog test passed on all the buildslaves
[07:37:49] -!- DaViruz has quit [Read error: Connection reset by peer]
[07:38:46] -!- Tom_itx has quit [Ping timeout: 245 seconds]
[07:41:11] -!- IceKiller has quit [Read error: Connection reset by peer]
[07:56:56] -!- racycle has quit [Quit: racycle]
[07:59:09] -!- vladimirek has quit [Remote host closed the connection]
[08:01:43] -!- tripod| has quit [Ping timeout: 264 seconds]
[08:10:20] -!- tmcw has quit [Ping timeout: 258 seconds]
[08:21:22] -!- tjb1 has quit [Quit: tjb1]
[08:27:54] -!- Tom_itx has quit [Ping timeout: 276 seconds]
[08:33:09] -!- tomate_ [tomate_!~quassel@2a01:e34:ee61:a5f0:9d8b:fe21:e768:85b3] has joined #linuxcnc-devel
[08:36:26] -!- stsydow has quit [Remote host closed the connection]
[08:37:04] Tom_sbc is now known as Tom_itx
[09:02:56] -!- fomox has quit [Ping timeout: 256 seconds]
[09:03:06] -!- rob_h [rob_h!~rob_h@027c16a0.bb.sky.com] has joined #linuxcnc-devel
[09:31:18] -!- MarkusBec has quit [Ping timeout: 264 seconds]
[09:32:33] <seb_kuzminsky> andypugh, jepler: reverting 606d8328f fixes the halui jogging problem, i have no idea why
[09:36:21] -!- Loetmichel has quit [*.net *.split]
[09:36:21] -!- tronwizard has quit [*.net *.split]
[09:36:21] -!- emel has quit [*.net *.split]
[09:36:21] -!- PCW has quit [*.net *.split]
[09:36:21] -!- uwe_mobile has quit [*.net *.split]
[09:36:21] -!- steves_logging has quit [*.net *.split]
[09:36:22] -!- pcw_home has quit [*.net *.split]
[09:36:22] -!- hewball has quit [*.net *.split]
[09:36:22] -!- hm2-buildmaster has quit [*.net *.split]
[09:42:38] -!- hm2-buildmaster [hm2-buildmaster!~hm2-build@174-16-204-191.hlrn.qwest.net] has joined #linuxcnc-devel
[09:43:51] -!- SAHarvey has quit [Ping timeout: 258 seconds]
[09:45:20] -!- pcw_home [pcw_home!~chatzilla@ip-66-80-167-54.sjc.megapath.net] has joined #linuxcnc-devel
[09:50:44] -!- L33TG33KG34R has quit [Ping timeout: 256 seconds]
[09:55:15] -!- b_b has quit [Changing host]
[10:00:23] -!- vladimirek [vladimirek!~vladimire@95.105.250.72] has joined #linuxcnc-devel
[10:02:25] -!- RagingComputer has quit [Ping timeout: 260 seconds]
[10:04:04] -!- andypugh [andypugh!88082144@gateway/web/freenode/ip.136.8.33.68] has joined #linuxcnc-devel
[10:04:51] <andypugh> Seb referenced commit 606d8328f but I can't find that in Gitweb. Maybe I don't know how to work Gitweb?
[10:11:00] -!- tmcw has quit [Ping timeout: 248 seconds]
[10:14:35] -!- PCW_ has quit [Remote host closed the connection]
[10:15:47] -!- dhoovie has quit [Ping timeout: 246 seconds]
[10:16:54] -!- mackerski has quit [Ping timeout: 264 seconds]
[10:16:54] mackerski_ is now known as mackerski
[10:20:01] -!- mhaberler [mhaberler!~mhaberler@extern-187.stiwoll.mah.priv.at] has joined #linuxcnc-devel
[10:26:24] -!- jst has quit [Quit: No Ping reply in 180 seconds.]
[10:30:43] -!- mackerski has quit [Quit: mackerski]
[10:35:11] -!- sumpfralle has quit [Quit: Leaving.]
[10:41:16] -!- fomox has quit [Ping timeout: 245 seconds]
[10:44:20] -!- dhoovie|2 has quit [Read error: Connection reset by peer]
[10:47:54] -!- sumpfralle has quit [Ping timeout: 245 seconds]
[10:56:58] -!- skunkworks has quit [Remote host closed the connection]
[11:06:00] Cylly is now known as Loetmichel
[11:11:15] -!- tmcw has quit [Ping timeout: 258 seconds]
[11:12:42] -!- mhaberler has quit [Quit: mhaberler]
[11:18:23] -!- toastyde1th has quit [Read error: Connection reset by peer]
[11:22:22] -!- Youdaman has quit [Client Quit]
[11:24:13] -!- SAHarvey has quit [Quit: SAHarvey]
[11:25:20] -!- V0idExp has quit [Quit: Leaving.]
[11:30:05] -!- mhaberler [mhaberler!~mhaberler@macbook.stiwoll.mah.priv.at] has joined #linuxcnc-devel
[11:30:22] -!- skunkworks [skunkworks!~skunkwork@68-115-41-210.static.eucl.wi.charter.com] has joined #linuxcnc-devel
[11:44:00] -!- zlog has quit [Remote host closed the connection]
[11:44:07] -!- Tom_itx has quit []
[11:46:44] -!- MarkusBec has quit [Ping timeout: 252 seconds]
[11:51:50] -!- ravenlock has quit [Quit: Leaving]
[11:53:23] -!- andypugh has quit [Ping timeout: 245 seconds]
[11:53:26] <jepler> andypugh: never use gitweb. just "git show 606d8328f"
[11:53:50] <jepler> andypugh: if you were trying to find it by going back through successive pages of history, it would take awhile: the change was in november 2011
[11:55:15] <skunkworks> wow - and no one had noticed it until now?
[12:01:09] <jepler> Your branch is behind 'origin/master' by 1389 commits, and can be fast-forwarded.
[12:01:17] <jepler> boy, I don't work on linuxcnc much anymore, do I...
[12:03:18] -!- skunkworks_ [skunkworks_!~skunkwork@68-115-41-210.static.eucl.wi.charter.com] has joined #linuxcnc-devel
[12:06:10] -!- skunkworks has quit [Ping timeout: 256 seconds]
[12:11:30] -!- tmcw has quit [Ping timeout: 264 seconds]
[12:19:05] <jepler> seb_kuzminsky: your test doesn't work for me
[12:19:11] <jepler> something's going wrong anyway
[12:19:14] <jepler> I think ths may be the real error?
[12:19:15] <jepler> + ./test-ui.py -ini /usr/src/linuxcnc/tests/halui-jogging/halui.ini
[12:19:15] <jepler> /usr/bin/env: python2: No such file or directory
[12:24:38] <jepler> I think this is the bug:
[12:24:38] <jepler> hal_bit_t bit, js;
[12:24:39] <jepler> ...
[12:24:42] <jepler> js = new_halui_data.joint_selected;
[12:25:47] <jepler> before the change that seb_kuzminsky found, the effective range of js was -128..127, plenty enough for all the joint numbers
[12:25:53] <jepler> after, as the comment says, the range is only two values
[12:29:20] <jepler> (or maybe the range was 0..255 .. either way)
[12:30:02] <KGB-linuxcnc> 03jepler 05v2.5_branch 33d4483 06linuxcnc 10src/emc/usr_intf/halui.cc * halui: don't represent joint number in a hal_bit_t!
[12:31:53] <jepler> seb_kuzminsky: I was going to merge that into master, but there's a conflict with the mdi stuff that I would have to merge at the same time, so I'm punting instead.
[13:09:26] -!- tomate_ has quit [Ping timeout: 256 seconds]
[13:11:53] -!- tmcw has quit [Ping timeout: 255 seconds]
[13:15:14] -!- zzolo has quit [Quit: zzolo]
[13:15:14] -!- ler_hydra has quit [Remote host closed the connection]
[13:20:32] -!- odogono has quit [Ping timeout: 252 seconds]
[13:24:56] -!- Wildhoney has quit [Ping timeout: 252 seconds]
[13:28:28] <cradek> seb_kuzminsky: another big area that's currently untestable and historically shaky is css and spindle control, m3/m4/m5/g43/g96. for example see 1113a06a65
[13:29:10] <cradek> dc970e6fb
[13:29:20] -!- tmcw has quit [Remote host closed the connection]
[13:29:28] <cradek> fdb6e630
[13:29:41] <cradek> c2bc8cf3
[13:31:26] <cradek> (IMO this is because the nml messages and the gcode don't really correspond very well, so you have to keep state in various places)
[13:41:52] -!- chillly has quit [Quit: Leaving]
[13:43:09] -!- Wildhoney has quit [Ping timeout: 276 seconds]
[13:49:08] -!- odogono has quit [Ping timeout: 252 seconds]
[13:50:16] -!- Valen has quit [Quit: Leaving.]
[13:52:07] <seb_kuzminsky> jepler: yeah i still owe that merge to master...
[13:52:44] <seb_kuzminsky> i wonder why the bug doesnt manifest in 2.5, only in master
[13:53:54] <seb_kuzminsky> i guess the compiler gets a lot of freedom about the size and behavior of a bool
[13:54:28] <seb_kuzminsky> one of those wonderful "undefined" sections of C
[13:55:18] <seb_kuzminsky> maybe thays why it fails on my machine and not on the buildbot?
[13:56:38] <seb_kuzminsky> cradek: yeah there's a lot of additional testing we could (should?) be doing
[14:12:07] -!- b_b has quit [Changing host]
[14:13:30] -!- mackerski has quit [Quit: mackerski]
[14:14:35] -!- holst has quit [Quit: Leaving]
[14:15:38] -!- Wildhoney has quit [Ping timeout: 258 seconds]
[14:19:12] -!- kwallace [kwallace!~kwallace@smb-225.sonnet.com] has joined #linuxcnc-devel
[14:24:11] -!- Wildhoney has quit [Ping timeout: 245 seconds]
[14:26:16] -!- fomox has quit [Ping timeout: 245 seconds]
[14:30:21] <jepler> seb_kuzminsky: I think it's because 606d8328f is not in v2.5_branch .. isn't it?
[14:30:37] <jepler> dunno why it would have succeeded on buildbot though
[14:31:37] <cradek> yes the breaking change is not in 2.5
[14:32:00] -!- mhaberler has quit [Quit: mhaberler]
[14:40:20] -!- mk0 has quit [Ping timeout: 248 seconds]
[14:41:34] -!- psha[work] has quit [Quit: Lost terminal]
[14:52:18] -!- Tom_itx has quit [Ping timeout: 264 seconds]
[14:52:42] -!- zlog has quit [Ping timeout: 276 seconds]
[15:33:03] -!- ve7it [ve7it!~LawrenceG@S01060014d19d0b68.pk.shawcable.net] has joined #linuxcnc-devel
[15:33:48] -!- vladimirek has quit [Remote host closed the connection]
[15:43:04] <seb_kuzminsky> oh, that would explain it...
[15:54:48] <seb_kuzminsky> it bugs me that the scripts/linuxcnc hides stdout and stderr of everything
[15:55:11] <seb_kuzminsky> it makes it hard to tell what happened during runtests on the buildbot
[15:55:26] -!- jfrmilner has quit [Quit: bye]
[15:56:30] <seb_kuzminsky> scripts/linuxcnc accepts -v for verbose and -d for debug, but neither exposes the stdout/stderr of the processes it spawns (task, the ui, etc)
[15:57:21] -!- MyDrupalJourney has quit [Quit: MyDrupalJourney]
[15:58:36] <seb_kuzminsky> those outputs go to /tmp/linuxcnc.{print,debug}.$RANDOM, and they're removed on exit
[15:58:41] * seb_kuzminsky has a sad
[15:59:09] <cradek> I think they're (sometimes) shown in a window if it crashes
[15:59:33] <cradek> users really never see stdout and stderr
[16:01:58] <seb_kuzminsky> yeah, on crash the files are copied to $HOME, and if $DISPLAY is set it's shown in a little gui window
[16:02:09] <seb_kuzminsky> neither of which is useful on the buildbot :-/
[16:02:28] <cradek> just fix -v or -d?
[16:03:50] <seb_kuzminsky> maybe make "-d" tee stdout and stderr? so the old behavior is unchanged, and it also shows up on the terminal?
[16:04:12] <cradek> sure, if you can
[16:04:15] <seb_kuzminsky> would you accept that in 2.5?
[16:04:23] <cradek> not sure how tee and exec>> will work together
[16:04:29] <seb_kuzminsky> it's not critical, but it would be nice
[16:04:33] <seb_kuzminsky> i'll try it
[16:04:52] <cradek> yes if we test that it doesn't break the gui thing, and review it carefully
[16:04:56] <seb_kuzminsky> exec doesnt mess with file descriptors, so it Should Work(tm)
[16:20:38] <seb_kuzminsky> regarding jepler's halui fix, i'm surprised that our compilers aren't warning about the overflow on line 1941:
http://git.linuxcnc.org/gitweb?p=linuxcnc.git;a=blob;f=src/emc/usr_intf/halui.cc;h=5633fa8939408b8ffcaac87799431e98fa124ffe;hb=33d4483ccf856fea23726c3bcdf1ab7c19b855dd#l1941
[16:21:03] <seb_kuzminsky> joint_selected is a hal_u32_t, js was a hal_bit_t
[16:21:55] <cradek> yeah, I built it this morning in disbelief, looking for a warning
[16:22:22] <seb_kuzminsky> maybe that crazy FIELD macro that halui is using? maybe the crazy template stuff?
[16:24:50] <cradek> halui.o is built with g++ -Os -g -Wall
[16:25:51] <cradek> changing to -O6 still does not make a warning
[16:27:28] <seb_kuzminsky> -O is just for the optimizer, shouldnt have anything to do with warnings
[16:27:53] <cradek> -O level affects warnings *a lot*
[16:28:16] <cradek> -O0 disables a whole lot of warnings
[16:28:18] <seb_kuzminsky> gcc never ceases to surprise me
[16:28:55] <cradek> it's no surprise that more optimization analysis means more weirdnesses are discovered
[16:29:18] <cradek> and issuing warnings is always optional :-/
[16:30:17] <seb_kuzminsky> i built it with clang, which also did not warn about the overflow
[16:30:58] <cradek> http://gcc.gnu.org/onlinedocs/gcc-4.4.7/gcc/Warning-Options.html#Warning-Options
[16:31:04] <cradek> on this page, search for optimiz
[16:31:41] <seb_kuzminsky> yeah, sure enough
[16:32:00] -!- tandoori has quit [Read error: Connection reset by peer]
[16:32:04] <cradek> wonder which of these warnings we expect...?
[16:33:51] <cradek> -Wconversion?
[16:34:14] bluefr0g is now known as redfr0g
[16:35:39] <seb_kuzminsky> -Wconversion does emit 4 new warnings
[16:35:40] <cradek> I'm not sure it should trigger a warning. using an integer type in a boolean sense is perfectly normal in C
[16:37:13] <seb_kuzminsky> emc/usr_intf/halui.cc:1940:25: warning: conversion to ‘hal_bit_t {aka unsigned char}’ from ‘VALUE::field<volatile unsigned int>::type {aka unsigned int}’ may alter its value [-Wconversion]
[16:37:13] -!- r00t4rd3d_ has quit [Client Quit]
[16:37:27] <seb_kuzminsky> that looks like what we want
[16:37:31] <cradek> yep
[16:37:40] <cradek> and how many UNwanted ones do you get?
[16:37:56] -!- r00t4rd3d has quit [Disconnected by services]
[16:38:00] -!- r00t4rd3d_ has quit [Client Quit]
[16:38:15] <seb_kuzminsky> there are 8 conversion warnings in halui
[16:38:30] <seb_kuzminsky> they all look valuable
[16:38:45] <cradek> I meant in the whole source tree...
[16:38:49] <seb_kuzminsky> oh
[16:38:51] <seb_kuzminsky> i dont know
[16:39:13] <seb_kuzminsky> i think in general i'd rather have more warnings than fewer... but i realize not everyone feels this way
[16:40:32] <seb_kuzminsky> heh, 2228 warnings with -Wconversion
[16:40:50] -!- morfic has quit [Ping timeout: 260 seconds]
[16:41:14] <cradek> neither 0 nor 2228 warnings would have pointed us to this problem
[16:41:39] <seb_kuzminsky> vs 149 warnings with just -Wall
[16:41:53] <cradek> wow I had no idea there were that many
[16:42:04] <seb_kuzminsky> i'd love to fix all those warnings and then turn on -Werror
[16:43:00] <seb_kuzminsky> many of them are a warning about redefining FORTIFY_SOURCE
[16:43:16] <cradek> I would sure support trying to get to 0 warnings in master
[16:43:23] <seb_kuzminsky> cool
[16:43:51] <seb_kuzminsky> warnings are often harmless, but sometimes they point out sneaky bugs like this one
[16:43:52] <cradek> -Werror on buildbot would be fine, not sure about in general. different compilers give different warnings, and we might be shooting at others' feet.
[16:44:34] <cradek> I don't have particularly strong feelings about it
[16:45:22] <seb_kuzminsky> that FORTIFY thing came from here, one of my favourite commits:
http://git.linuxcnc.org/gitweb?p=linuxcnc.git;a=commitdiff;h=a99689b9f6c5d80797f47b9d68f28f9af84e2f63
[16:45:24] <cradek> also (like with static analysis) you sometimes end up with sillyfied code just to make the warnings shut up
[16:45:41] <seb_kuzminsky> yeah, that's a downside for sure
[16:46:00] <seb_kuzminsky> also, since many warnings are spurious, it can be a waste of time
[16:46:06] <cradek> yeah
[16:46:21] <seb_kuzminsky> i think it's probably time to stop defining FORTIFY_SOURCE in our compiler command line
[16:46:30] <seb_kuzminsky> i think it's on by default on all modern versions of gcc
[16:46:33] <cradek> although you do tend to eliminate the "oh ignore them they're just noise" opinion, which is an improvement
[16:47:06] <cradek> i show you how use not redefine FOURTIFY
[16:48:12] <seb_kuzminsky> hardy doesnt run with fortify by default
[16:48:33] <seb_kuzminsky> lucid neither
[16:48:35] <seb_kuzminsky> :-/
[16:48:37] <cradek> hm
[16:48:48] <seb_kuzminsky> by "modern" i mean precise and newer, apparently
[16:49:32] -!- MarkusBec has quit [Ping timeout: 252 seconds]
[16:54:15] -!- tripod| has quit [Ping timeout: 276 seconds]
[16:54:42] <seb_kuzminsky> our makefiles explicitly set FORTIFY_SOURCE to 0, that's what hardy and lucid do by default
[16:55:02] <seb_kuzminsky> but on precise the default is FORTIFY_SOURCE=2 (i think) and it complains about it
[16:55:25] <seb_kuzminsky> i think FORTIFY_SOURCE is a good idea and should probably always be on (unless it's broken in Hardy & Lucid, i guess)
[16:56:20] <seb_kuzminsky> maybe just removing the explicit setting of it would get us what we want? hardy and lucid keep it off and precise gets it on
[16:58:17] <seb_kuzminsky> without -D_FORTIFY_SOURCE=0 i get 19 warnings on precise (instead of 149)
[16:58:35] <cradek> cool
[16:58:44] <seb_kuzminsky> wonder if runtests still passes?
[17:05:54] -!- V0idExp has quit [Quit: Leaving.]
[17:07:17] <seb_kuzminsky> passes on precise sim
[17:10:33] -!- ink has quit [Disconnected by services]
[17:13:20] -!- V0idExp has quit [Client Quit]
[17:14:21] <seb_kuzminsky> hardy and lucid both get -D_FORTIFY_SOURCE=0 from "rtai-config --module-cflags", and i think that's fine
[17:14:38] <cradek> cool, thanks for figuring it out
[17:15:37] -!- andypugh [andypugh!~andy2@cpc16-basl9-2-0-cust685.20-1.cable.virginmedia.com] has joined #linuxcnc-devel
[17:21:11] -!- mhaberler [mhaberler!~mhaberler@macbook.stiwoll.mah.priv.at] has joined #linuxcnc-devel
[17:25:39] <seb_kuzminsky> passes on lucid-rtai without it
[17:27:45] <seb_kuzminsky> and on hardy-rtai
[17:28:05] <andypugh> What is the path for jog bugfix into Master? Next merge?
[17:28:17] <seb_kuzminsky> i'm gonna remove it in 2.5, yeah boo?
[17:28:23] <seb_kuzminsky> andypugh: yeah
[17:28:38] <seb_kuzminsky> i need to merge the mdi queue fix first, that one's not trivial
[17:29:03] <andypugh> OK, jepler's message hinted there were dependencies on the mdi fix, though that might just have been on his system.
[17:29:32] <andypugh> Ah, OK.
[17:29:43] <seb_kuzminsky> the two fixes are not dependent on each other, but since the mdi fix is before the jog fix, it needs to be merged first
[17:30:02] <andypugh> <Unless we rewrite git history>
[17:30:10] <seb_kuzminsky> yeah, true
[17:30:20] <seb_kuzminsky> but the mdi queue fix needs to be merged anyway, so it doesn't seem worth it
[17:30:38] <andypugh> An alternative bugfix would be to re-use the perfectly good "joint" variable rather than the used-once "js" :-)
[17:31:07] -!- mackerski has quit [Quit: mackerski]
[17:31:52] <andypugh> And how about fixing the indentation in halui.cc at the same time? The joint loop is nicely hidden.
[17:32:27] <andypugh> Nothing like a huge whitespace-only commit to confus things.
[17:32:44] -!- tandoori has quit [Changing host]
[17:34:30] <seb_kuzminsky> awesome, in 2.5 we force FORTIFY_SOURCE=0, in master it's FORTIFY_SOURCE=2
[17:34:48] <seb_kuzminsky> i think we should remove the -D on both, and let the compiler and/or rtai-config choose
[17:36:56] <andypugh> Does it ever make sense for a component to have a loopback pin pair? For example an imput that is there in all modes, and an output to match that input in a subset of modes.
[17:38:17] <seb_kuzminsky> might make sense in some circumstance, i dont know
[17:49:03] -!- tmcw has quit [Remote host closed the connection]
[17:53:46] -!- psha [psha!~psha@213.208.162.67] has joined #linuxcnc-devel
[17:59:18] -!- dway has quit [Quit: dway]
[17:59:34] -!- vladimirek [vladimirek!~vladimire@95.105.250.72] has joined #linuxcnc-devel
[18:16:44] <seb_kuzminsky> how embarrassing to have to admit my blunder on the rtai list
[18:19:41] -!- phillipadsmith has quit [Ping timeout: 255 seconds]
[18:25:46] <cradek> looks like a minor thing
[18:26:18] <cradek> or is it expected not to work? IMO smp kernel on a UP machine is a very important test case
[18:27:53] -!- XiXora has quit [Read error: Operation timed out]
[18:27:56] -!- mattswe has quit [Ping timeout: 245 seconds]
[18:29:00] -!- beawesomeinstead has quit [Ping timeout: 240 seconds]
[18:29:40] <cradek> whoah, serious language obstacles, but I think it's a store and not their house, they're open every day until 7, and the stove works but is not hooked up to gas now.
[18:29:49] <cradek> haha
[18:29:53] <cradek> that's not the right channel
[18:31:13] <andypugh> This the Cistene Chapel stove you are talking about?
[18:31:35] <cradek> unless it's being sold on craigslist, no
[18:32:53] <cradek> we were making fun of the white smoke coming out of the chimney of the fast food place across the street this morning
[18:35:06] <andypugh> It's a fairly cool tradition, isn't it?
[18:35:37] <cradek> can't decide if it's cool or silly - a mix of both I guess
[18:36:03] <cradek> it sure gets a lot of attention
[18:36:16] <andypugh> It would be silly if they thought it up now. But because they have done it that way for centuries, its good.
[18:37:58] -!- V0idExp has quit [Quit: Leaving.]
[18:38:27] <skunkworks_> spooky oddity.. I asked my wife - who grew up in a catholic familly - if the pope always dies in office - or do they ever step down? this was the weekend before the pope stepped down.
[18:38:48] <cradek> apparently it's been done before
[18:38:53] <skunkworks_> I need to figure out how to harnes my power.
[18:38:59] <skunkworks_> yes - a few centurys ago
[18:39:27] <skunkworks_> centuries
[18:40:22] <andypugh> Last resignation was 1415.
[18:42:14] -!- servos4ever has quit [Ping timeout: 246 seconds]
[18:53:27] -!- nunya has quit [Quit: Leaving]
[18:55:50] Guest64639 is now known as fragalot
[18:55:57] -!- fragalot has quit [Changing host]
[19:06:51] -!- sumpfralle has quit [Ping timeout: 260 seconds]
[19:08:39] -!- sumpfralle1 has quit [Read error: Connection reset by peer]
[19:14:28] -!- sumpfralle has quit [Ping timeout: 248 seconds]
[19:18:08] -!- zzolo has quit [Read error: Connection reset by peer]
[19:22:39] -!- Chendy has quit [Client Quit]
[19:24:40] -!- Thetawaves_ has quit [Quit: This computer has gone to sleep]
[19:25:24] -!- b_b has quit [Changing host]
[19:29:26] <skunkworks_> does anyone know where the stepgen waveform diagrams are in the 2.5 docs?
[19:30:18] <skunkworks_> is this it?
[19:30:19] <skunkworks_> http://linuxcnc.org/docs/html/hal/rtcomps.html
[19:32:13] <seb_kuzminsky> that looks like it
[19:32:23] <skunkworks_> Thanks
[19:32:24] <seb_kuzminsky> was it always in color? i recall them as being black & white
[19:32:52] -!- ravenlock has quit [Quit: Leaving]
[19:33:12] <cradek> I think they're ascii art in the manpage
[19:33:23] <skunkworks_> do you linuxcnc in black and white?
[19:33:40] <seb_kuzminsky> this must be a meme i dont know
[19:33:41] <jepler> they were black and white here:
http://linuxcnc.org/docs/2.4/html/hal_rtcomps.html#fig:Stepgen-Block-Diag
[19:33:47] <jepler> somebody's improved them in the last 10 years?
[19:34:09] <seb_kuzminsky> tiss!
[19:34:36] <cradek> are they even right? 7 and 10 look weird
[19:34:49] <seb_kuzminsky> killjoy!
[19:34:51] <seb_kuzminsky> they're in color now!
[19:34:52] <seb_kuzminsky> better!
[19:35:09] <seb_kuzminsky> we have three copies of the stepgen block diagram
[19:35:21] <seb_kuzminsky> an old eps in b&w, and a new svg + png in color added by tiss
[19:35:42] <jepler> they look the same in 2.4 and 2.5 so .. that's all I know
[19:35:58] <cradek> 1. FF2 is not currently implemented, but it will be added. Consider this note a "FIXME" for the code
[19:35:59] <jepler> (step types 7 and 10)
[19:41:34] -!- b_b has quit [Changing host]
[19:49:30] -!- phantoxeD has quit [Ping timeout: 240 seconds]
[19:51:42] -!- syyl has quit [Ping timeout: 264 seconds]
[20:01:30] -!- mhaberler has quit [Ping timeout: 240 seconds]
[20:06:06] <seb_kuzminsky> public-facing documentation is a funny place for bug tracking
[20:06:34] <cradek> and also it was implemented like a decade ago
[20:12:21] -!- skunkworks_ has quit [Read error: Connection reset by peer]
[20:21:41] -!- snkashis has quit [Quit: Leaving.]
[20:25:32] -!- tjb1 has quit [Ping timeout: 258 seconds]
[20:35:37] -!- r00t4rd3d has quit [Ping timeout: 256 seconds]
[20:39:54] -!- skunkworks [skunkworks!~chatzilla@str-broadband-ccmts-ws-26.dsl.airstreamcomm.net] has joined #linuxcnc-devel
[20:45:02] -!- uncrtnmind has quit [Quit: ["Textual IRC Client: www.textualapp.com"]]
[20:53:26] -!- JT-Shop has quit [Ping timeout: 252 seconds]
[20:54:12] -!- jthornton has quit [Ping timeout: 248 seconds]
[20:54:22] -!- FinboySlick has quit [Quit: Leaving.]
[21:03:54] -!- V0idExp has quit [Quit: Leaving.]
[21:16:03] -!- AphelionZ has quit [Quit: Page closed]
[21:18:57] -!- psha has quit [Quit: Lost terminal]
[21:23:44] -!- jthornton [jthornton!~john@162.72.171.228] has joined #linuxcnc-devel
[21:24:11] -!- zzolo has quit [Quit: zzolo]
[21:24:36] -!- Wildhoney has quit [Ping timeout: 252 seconds]
[21:29:05] -!- motioncontrol has quit [Quit: Sto andando via]
[21:29:10] -!- syyl_ has quit [Quit: Leaving]
[21:49:00] -!- sumpfralle1 has quit [Ping timeout: 240 seconds]
[21:50:47] -!- tjb1 has quit [Quit: tjb1]
[22:00:32] <andypugh> RTAPI_MP_ARRAY_STRING was allowing me to define a modparam for a component perfectly happily, and now it isn't any more.
[22:00:58] <andypugh> Any experience of weird side-effects of other errers that can do that?
[22:02:07] <andypugh> I am getting an "unknown parameter" error, but it used to work as expected.
[22:04:36] <seb_kuzminsky> what did you change?
[22:06:20] <andypugh> Unfortunately i can't remember.
[22:06:51] <andypugh> I did add a third (integer) modparam. But commenting the others out doesn't help
[22:07:41] <seb_kuzminsky> what platform are you on? kernel & distro
[22:08:25] <andypugh> Good point, I might have changed branches via cherry-pick..
[22:08:30] -!- Loetmichel has quit [Ping timeout: 264 seconds]
[22:09:02] <seb_kuzminsky> errr, cherrypicking doesnt change branches
[22:09:41] <seb_kuzminsky> i did a bugfix in 2.5 recently to fix modparams on linux 3.5
[22:10:07] <seb_kuzminsky> it was a header file problem in rtapi, and it affected smartserial
[22:10:19] -!- r00t4rd3d_ has quit [Quit: Leaving]
[22:10:56] <andypugh> I mean, changed the branch that my new stuff is based on. I was using this component to check the jogging bug (as it lets me use a keyboard on my desk to jog with). That meant having the new thing in both branches.
[22:11:22] <seb_kuzminsky> oh i see
[22:11:50] <andypugh> I like to make things harder than necessary, clearly.
[22:13:33] <andypugh> I seem to be in a branch of of master
[22:14:03] <andypugh> But I think I always was.
[22:15:26] -!- DJ9DJ has quit [Quit: bye]
[22:17:50] <andypugh> seb_kuzminsky: I am pretty sure it isn't to do with you, it worked yesterday
[22:18:14] <seb_kuzminsky> that's what i like to hear
[22:20:58] <andypugh> OK, and now it works again. That's rather random
[22:22:09] <andypugh> No code changes.
[22:24:00] -!- chillly has quit [Quit: Leaving]
[22:24:18] <andypugh> It takes a size= and a names=. I tried just names, it complained that size and names had to agree on how many instances. I then tried just "sizes" and it said that "sizes" was wrong. Then I tried "size" and it was happy. And, scrolling back, it isn't the obvious explanation that I had it wrong ("sizes") before.
[22:25:54] ink is now known as ink_
[22:33:12] -!- micges [micges!~micges@adda83.neoplus.adsl.tpnet.pl] has joined #linuxcnc-devel
[22:34:21] <andypugh> http://youtu.be/rLr2G-1WPTI
[22:34:38] <andypugh> (Actually, I meant that for the other channel)
[22:48:26] -!- micges has quit [Quit: Leaving]
[22:50:30] -!- ink_ has quit [Ping timeout: 264 seconds]
[22:50:45] -!- adb [adb!~IonMoldom@178.211.237.94] has joined #linuxcnc-devel
[22:51:54] -!- rumapper has quit [Quit: Page closed]
[22:56:17] -!- V0idExp has quit [Quit: Leaving.]
[23:18:27] -!- odogono has quit [Quit: odogono]
[23:24:13] -!- joe9 [joe9!~joe9@c-24-98-97-215.hsd1.ga.comcast.net] has joined #linuxcnc-devel
[23:33:20] -!- fomox has quit [Ping timeout: 255 seconds]
[23:40:48] -!- fomox has quit [Remote host closed the connection]
[23:46:02] -!- Dupe has quit [Quit: Leaving]
[23:48:02] -!- tmcw has quit [Remote host closed the connection]