00:00:07 (how are you creating them) 00:01:05 emacs :D 00:01:34 emacs? apple? 00:01:39 yeah :) 00:02:14 really - would you mind explaining? (you don't have to) 00:02:17 oh .. youre not identified to the server ? :/ 00:02:30 why ? 00:03:05 yes, I used some gcode file a friend gave me 00:03:17 no I am not. I can see what you are typing but can't respond 00:03:27 oh.. ok 00:04:16 I seen some job you done in frappr photos 00:04:26 its nice :) 00:05:02 the ball? that was done with lermans interp and a little right angle trig ;) 00:05:07 yeah 00:05:41 one of these days I will make finish one so that the ball is trapped 00:05:51 cool 00:06:02 its very nice 00:06:27 It was our first full 3 axis machine - had to play a litte ;) 00:07:18 I'm thinking to add a tilt to Z axis of mine cnc 00:07:29 but its a toy .. 00:07:44 I dont wnat to spend much time on it 00:08:16 we need to get a rotory axis on ours to play. 00:08:18 night 00:08:24 k4ts has quit 00:08:46 that software looks neet 00:08:53 could be funny :) 00:11:40 btw, my first machine gave me a lots of satisfactions, im going to write a small doc about it to help peoples who do not have to much money to start with an expensive machine 00:12:31 to be honest, it was really funny 00:14:07 I guessing you mean 'really fun'. as in it was a good time? 00:14:47 yeah, too 00:15:17 I forced it a bit .. 00:16:00 starting with a dremel and changing it with 1kw router 00:16:27 all depend on what peoples want to do 00:16:38 03alex_joni * 10emc2/src/ (configure configure.in): checking for sudo, also allows overrides like this: --with-sudo=/usr/bin/gksudo 00:16:47 I just think computer controlled multi axis machine is fun - we might just be odd 00:18:19 03alex_joni * 10emc2/scripts/emc.in: changed sudo with found by configure 00:20:17 http://timeguy.com/cradek-files/emc/control 00:20:24 oops 00:20:29 :) 00:20:39 done that before ;) 00:27:37 haha, three people immediately loaded that url 00:30:49 lol.. it's yummy 00:30:50 :D 00:32:14 cradek: im too :(°°° 00:32:27 did you logged my IP ? O:O 00:32:37 probably 00:32:43 aarrggh 00:32:46 :( 00:33:06 I admit I did it also ;) 00:33:27 oh.. ok, at least im not the only one 00:34:11 81.211.255.209 - - [28/Jan/2006:18:20:47 -0600] "GET /cradek-files/emc/control HTTP/1.1" 200 634 "-" "Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.8) Gecko/20060110 Debian/1.5.dfsg-4 Firefox/1.5" 00:35:10 ERC Version 3.0 $Revision: 00:35:11 1.477 $ (IRC client for Emacs) 00:35:17 :D 00:35:24 a bit of spam :P 00:36:36 cradek: what is apache ? 00:36:44 I like Boa 00:37:07 for embedded systems.. its incredible, fassst 00:37:25 and neat 00:44:36 Jymmm has joined #emc 00:59:38 hey Jymmm 00:59:45 hi 01:00:04 is cold there ? 01:00:49 http://weather.noaa.gov/weather/current/KSJC.html 01:01:13 oh 01:01:17 lax :) 01:01:32 Jymmm is now known as Red70sShow 01:01:32 Red70sShow is now known as Jymmm 01:01:34 oh.. no 01:01:53 Humid 01:01:59 but not cold 01:02:19 I was expecting the snow there 01:04:14 latest photo I did: 5 min ago http://digilander.libero.it/jackydgl0/chelin.jpg 01:04:41 damn kats.. eat eat .. E. 5 at day 01:04:45 :/ 01:15:19 Jacky^ : http://www.ooze.com/ooze13/cats.html 01:16:03 aaarrrghh 01:16:11 no :(°° 01:26:48 alex_joni_ has joined #emc 01:40:46 alex_joni_ has quit 01:44:40 night all 01:44:47 see ya 01:45:07 night alex_joni 01:45:38 im going to 01:45:43 see ya 01:45:46 night 01:46:01 Jacky^ is now known as Jacky^afk 02:24:09 jepler_ has joined #emc 02:31:46 jepler has quit 03:28:57 skunkworks has quit 03:38:44 ChanServ has quit 03:49:16 This nickname is owned by someone else 03:49:16 If this is your nickname, type /msg NickServ IDENTIFY 03:49:16 ChanServ has joined #emc 03:59:15 dmessier has quit 04:01:35 03cradek * 10emc2/scripts/emc.in: ugh, looks like we need this one too 06:21:33 Jymmm has quit 07:17:51 A-L-P-H-A has quit 07:42:43 A-L-P-H-A has joined #emc 07:44:51 Jymmm has joined #emc 07:55:15 Jymmm has quit 11:01:31 Jacky^afk is now known as Jacky^ 11:05:21 some engineer forgot the snow planning a building in Poland .. 11:06:01 as some other forgot the rain planning the Olympic Stadium - Montreal 11:06:58 we are in good hands 11:07:22 morning .. 11:25:32 03alex_joni * 10emc2/docs/AUTHORS: added jepler to the list of authors 11:26:22 03alex_joni * 10emc2/docs/NEWS: added latest bugfix 11:26:52 03alex_joni * 10emc2/debian/changelog: added latest bugfix 11:46:35 03alex_joni * 10emc2/directory.map: fixed typos, added some more information 11:56:11 something's borked in the makefile.. got 187 levels of recursion before i killed it.. 11:57:34 robin_sz has joined #emc 11:57:39 meep? 11:57:46 hi robin_sz :) 11:58:02 eveny 11:58:07 er, i was doing make tgz btw 11:58:16 seen the latest news from Poland ? 11:58:17 * fenn talks to nobody in particular 11:58:34 robin_sz: the engineer forgot the snow .. 11:58:40 oopsy 11:58:49 he sayd: damn it ! the snow .. I forgot it :/ 11:59:13 I presume this is yet another flat building? 11:59:41 it seems square on the top 12:00:05 have you noticed how swiss chalets have very sloping roofs? 12:00:26 nope 12:00:33 they do 12:00:42 prevents snow build up 12:01:55 hello 12:01:58 hi robin 12:02:01 hi alex_joni 12:02:05 gday 12:02:26 * robin_sz prepares to do a debian install on his laptop 12:02:39 wow 12:02:50 robin_sz: using sarge installer ? 12:02:55 I hope it leaves my redhat and windows installs alone 12:02:57 yeah 12:03:00 good 12:03:01 net install 12:03:04 yeah 12:03:44 "A catastrophe on this scale has never happened in Poland." ??? poland has been blasted to hell and back more than any other country 12:05:18 the problem is when an engineer plan without knowledge .. 12:05:33 and forgot some important 'thing' 12:05:47 as the rain, snow, wind 12:05:59 later .. 12:06:08 Jacky^ is now known as Jacky^lunch 12:06:32 * fenn wonders when people will learn to use geodesic domes.. 12:07:02 fenn: still ahving problems with make tgz? 12:07:18 I had the same issue, because CVS didn't checkout the latest sourcecode 12:07:25 and a hal/ dir was missing 12:07:36 hal/user_comps 12:08:03 robin_sz: I installed a sarge on a machine lately, works great 12:08:24 robin_sz: use xfce4, I love that.. 12:09:35 ok rebooting ... :) 12:09:40 robin_sz has quit 12:09:41 alex_joni: thanks that was the problem.. needed a -dP 12:09:51 weird symptom 12:10:15 yup 12:10:46 i think it tries to do (cd user_comps; make clean) but then its still in the same directory 12:11:04 yup 12:11:16 well.. at least it cleans up thoroughly 12:11:24 lol 12:11:40 heh maybe it should shred the disk after its done 12:11:54 I think my cron script doesn't do cvs up -dP properly 12:11:55 :( 12:13:07 pserver has been screwy for a long time 12:13:44 hrm.. i forget which method can accept a password 12:13:54 I don't use pserver 12:14:03 maybe its ext then 12:14:11 anyways, I didn't have -dP there ;) 12:14:15 ah 13:12:19 Jacky^lunch is now known as Jacky^ 13:24:26 alex_joni: already tried Ion3 ? 13:24:44 the window manager 13:25:13 I suggest to use it especially in laptop .. 13:25:59 just spending few days to remember keys combinations, youll be happy Forever ! :D 13:26:57 I was using kde in my 10" laptop, I was tired to resize windows and use the mouse .. 13:27:12 my right hand was tired :P 13:27:31 floating workspace is coool, try it 13:28:01 and tell me if you like it 13:29:04 for developers its the best wm 13:30:22 I had same difficult as in emacs starting with it, to remember sequence keys 13:30:39 but after a bit .. :D 13:31:19 right hand stay on keyboard 13:42:22 Jacky^: remember I don't have a mouse on my laptop 13:42:34 so I don't need to worry about moving my right hand 13:42:34 :D 13:43:11 http://digilander.libero.it/jackydgl0/desktop.png 13:44:49 thats an Intelligent window manager ! :P 13:50:01 i'm a ion fanatic 13:50:03 heh 13:57:46 can't see anything inteligent there.. 13:58:21 why you should resize a window ? 14:00:14 because I want sometimes to see how text output looks over a smaller terminal? 14:00:20 or the other way around? 14:02:57 I usually ahev a few windows open 14:03:15 say I have one for editing, and one for compiling, and one for halcmd commands 14:03:27 the compile and the halcmd commands, need to be small usually 14:03:38 and the editing window needs to be as large as possible 14:03:50 but once in a while, there might be a compile problem, 14:03:58 then I need the compile window to be large 14:04:13 but only so long as to read what's wrong 14:05:03 * fenn has 22 windows open on this desktop (out of 4 desktops) 14:05:07 you can always switch to full screnn with a simple meta+key 14:05:15 screen 14:05:54 for the rest, you no need to worry about resize and positioning anything 14:05:59 try it .. 14:06:27 going out with a friend to take a drink :P 14:06:28 can't say I'm a fan of software with a small userbase 14:06:31 later 14:06:36 later 14:06:43 Jacky^ is now known as Jacky^afk 14:09:01 wish my keyboard had a scroll wheel 14:11:51 right in that little empty space between the enter and up arrow 14:12:02 oh well 14:42:45 add one 14:44:25 thinking about it.. been reading about avr microcontrollers a lot lately 14:44:38 seems that adding a usb scroll wheel and some blinkenlights would be easy 14:51:04 har har 14:51:10 SWP_Away is now known as SWPadnos 15:00:11 no? 15:00:34 well at worst i could chop up a mouse 15:06:16 it's probably not that bad, if you find a library for the keyboard serial protocol 15:07:35 you mean mouse protocol? 15:08:56 either 15:09:30 rayh has joined #emc 15:09:35 I misse dthe "usb" in your previous statement though 15:09:38 missed the 15:10:02 you'd want one of the USB microcontrollers for that (AT42xxx, I think) 15:10:27 Hi Ray - just in time for breakfast 15:10:51 oops - AT43xx 15:11:48 ok - bavk in a bit 15:11:51 back 15:11:53 that is 15:11:59 SWPadnos is now known as SWP_Away 15:12:00 Hi Steven, guys 15:12:15 morning 15:12:29 You get any snow fenn 15:13:30 no just lots of rain 15:15:15 logger_aj: bookmark 15:15:16 See http://solaris.cs.utt.ro/irc/irc.freenode.net:6667/emc/2006-01-29#T15-15-15 15:17:40 Jacky^afk has quit 15:25:33 I'd like some comment on using halconfig. If you've got a working emc2 and time. 15:30:12 cncuser has quit 15:34:09 halconfig says couldn't execute "bin/halcmd": no such file or directory 15:34:14 when i click show all 15:34:55 i've seen it work before though.. 15:35:22 Are you starting halconfig from emc2 15:35:53 from tkemc scripts -> hal_show 15:36:34 That shouldn't be there at all. 15:37:35 It means that you've got a file named emc2/scripts/hal_show.tcl 15:37:47 Is this a recent emc2? 15:38:38 i just updated.. it had a bunch of crap still in there it didnt expect 15:39:00 Jacky^ has joined #emc 15:39:10 er, hal_show isn't halconfig then huh.. 15:40:12 no. I don't remember much about halshow but it was a very early effort. 15:40:47 Is this Debian? 15:41:16 no it's a well-hacked fedora 2 15:43:34 Ah. You have the axis interface running? 15:43:47 or laying around? 15:44:03 sure.. recompiling a fresh checkout of emc2.. hang on 15:45:07 k 15:46:31 You will need to link the bwidget directory from the axis source to /usr/lib/bwidget 15:46:44 or find a bwidget.rpm 15:46:53 eh, what? 15:50:28 bwidget is a tool library extension to tk widgets. 15:55:41 so i need to use axis to be able to use halconfig? (have some axis compilation errors) 15:56:18 acemi has joined #emc 15:56:47 No you just need to find the bwidget tools in the axis source 15:58:36 here it is axis-20051120/thirdparty/bwidget 15:59:12 link or drag that directory into /usr/lib 16:05:37 fenn: What problems are you having building axis? 16:05:39 jepler_ is now known as jepler 16:06:12 fenn: 1.1.1 or snapshots? 16:06:48 hi all 16:07:11 cd axis-20051120/; sudo env EMC2ROOT=/home/blipkowi/emc2/ python setup.py install 16:07:32 extensions/emcmodule.cc:395: error: 'class EMC_STAT' has no member named 'logFile' 16:07:41 and so on for all the member of EMC_STAT 16:08:12 i havent tried anything yet 16:09:09 fenn: The environment variable is EMCROOT even for emc2 16:09:26 fenn: Is that error on line 395 the first error? 16:09:32 same thing with EMCROOT 16:10:32 SWP_Away is now known as SWPadnos 16:10:47 jepler: yep, have a look: http://pastebin.com/528851 16:11:28 uf that chopped up the word wrapping pretty bad 16:11:43 have you done a full make clean / make for emc2? 16:12:05 yeah this is a fresh checkout 16:12:10 o 16:12:12 ok 16:12:51 fenn: why axis-20051120? That's pretty old. 16:13:17 fenn: you'd be much better off with axis-1.1.1 16:13:27 right - and the emc status struct has changed since then, I think 16:13:28 hrm i thought i just downloaded axis-latest 16:13:41 it's way past my bedtime but i'm up anyway 16:13:58 these log variables were removed in emc2 16:14:03 we fixed axis on 11/17 16:14:05 right 16:14:11 er, 11/27 16:14:15 http://unpy.net/cgi-bin/viewcvs.cgi/axis/extensions/emcmodule.cc.diff?r1=1.12;r2=1.13;f=h 16:14:19 axis changed the numbering scheme it seems 16:14:33 so axis-2 (tab complete) went to the wrong directory 16:14:42 oh 16:15:03 just delete that old tree 16:15:39 the "release of the minute" is numbered by patchset (-psNNN) 16:16:58 rayh: fyi axis now installs bwidget in /usr/share/axis/tcl/bwidget/ 16:17:25 Okay. 16:19:07 it looks like there's still no fedora bwidget package even for fedore core 4 16:19:36 too bad, I'd like to stop shipping bwidget with axis but it seems only debian has a package available. 16:20:34 So the quick answer for halconfig is to test for it's existence in /usr/share/axis/tcl/bwidget/ 16:21:11 check for /usr/lib/bwidget (the package version) first, then check /usr/share/axis 16:21:33 so if the user has their own bwidget installed it should use that one, right? 16:22:25 the details are fuzzy but cncuser had some problem on his puppy distribution. It has a version of bwidget installed, but it's not incompatible to the one in axis. 16:22:45 Yep. 16:22:53 er, not compatible 16:23:09 heh - damn, you cut off my sarcastic question ;) 16:23:10 ouch. 16:23:13 I fixed it in axis by using the version number in 'package require' 16:23:25 dan_falck has joined #emc 16:23:34 Hi Dan. 16:23:38 r.tk.call("package", "require", "BWidget", "1.7") 16:23:43 hi ray 16:24:46 adding the directory where AXIS' copy of bwidget would reside, then using 'package require BWidget 1.7' should be safe 16:26:35 jmkasunich has joined #emc 16:28:16 hiya jmk - long time no see 16:28:59 only a week-ish 16:29:17 I've had a cold, so it seems like longer, I guess ;) 16:34:33 rayh: what happens if you do "scripts/realtime start" and then tcl/bin/halconfig.tcl ? 16:35:21 it bombs here 16:35:30 you'll see halconfig with 5 nodes in the tree. 16:35:45 If realtime is not running it asks if you want to do that. 16:35:50 adeos isn't a module on my system so i guess scripts/realtime status fails? 16:36:14 I would issue "sudo tcl/bin/halconfig.tcl" 16:36:14 but it works if you start halconfig with realtime stopped 16:36:38 it works here ... 16:36:44 It should see that no hal is running and ask 16:36:48 I just see nothing in the tree 16:37:31 (except of course a couple of halcmd comps) 16:40:21 interesting - double-click now works for selecting a config in setupconfig 16:40:36 I just started with nothing HAL ish running and got a popup. 16:42:15 wish is hogging 90% of the cpu 16:42:28 roltek has joined #emc 16:43:59 is there an easy way to debug tcl? 16:45:04 odd - wish isn't even on the list in top, unless I do something in halconfig 16:45:39 does halconfig poll halcmd for screen updates? that might make it hog cpu 16:45:48 but it goes back to 0 when I'm done 16:45:58 it does, but on a scale of seconds 16:46:39 it worked before, now there's a blank window that appears until i kill it 16:47:21 but now it's not hogging cpu.. 16:47:47 so, this dovetails nicely (relatively) into a discussion Ray and I had this week - how to make a program that's more efficient for variable watches than halcmd ... 16:48:07 er, nevermind that last comment about it not hogging cpu 16:56:36 since it looks like ray has some more victims to toy with now, i'm going to get some sleep 16:56:44 night 16:56:45 * fenn goes to bed 16:59:01 skunkworks has joined #emc 17:02:22 * rayh reads back 17:04:21 * SWPadnos waits quietly 17:08:58 okay. 17:09:28 not sure what's happening with fenn - things seem sane here 17:09:34 (the computers, at least) 17:09:53 Yea I can't understand his problem either. 17:42:57 Jymmm has joined #emc 17:47:28 dan_falck has quit 17:58:32 jepler: are you here? 17:58:42 jmkasunich: yeah, what's up? 17:58:55 I just ran the revised setupconfig.tcl 17:58:59 very nice! 17:59:02 oh 17:59:04 I noticed something tho 17:59:18 what's that? 17:59:37 when you get to the screen in the "new config" sequence where you enter a name, there are some messages echoed to the shell window 17:59:50 button_entry_change {.main.f1.f2.next >} new_name . . op 17:59:56 oh yeah those are probably just junk 18:00:10 are they going to stderr? 18:00:20 maybe, how can I tell? 18:00:48 I suppose I could redirect stdout to /dev/null and see if they still happen 18:00:52 did you start setupconfig by running emc without an arguemnt? 18:01:17 if they're going to stderr they're harmless, but if they're going to stdout they could cause problems 18:01:20 no, I started it directly from a prompt 18:01:22 right 18:01:41 apparently stderr 18:01:56 I restarted with >/dev/null, and they still appear 18:02:00 are they real errors? 18:02:05 no, those are debug messages for me 18:02:18 for when I was trying to make sure the "Next >" button was greyed out when no new config file name had been entered 18:02:29 so they will go away eventually, good 18:03:07 btw, what do you think about merging the first two screens of the "new" process 18:03:49 "you have chosen" "the next few screens will walk you thru" blank line "enter a name" 18:04:02 sounds fine to me 18:04:53 03jepler * 10emc2/tcl/bin/setupconfig.tcl: remove debugging message 18:05:21 another question 18:05:39 what about the first screen "config/quit/run" 18:06:07 should run simply be added as a choice to the config menu screen 18:06:18 and cancel on that screen be changed to quit? 18:07:19 there are two levels of "main menu" right now, and perhaps it should only be one 18:07:40 the "choose a config and run it" behavior makes sense when the user just runs the emc2 script without an argument 18:07:55 having a "back" or "cancel" from the first screen the user sees is a bit odd 18:09:12 true 18:09:48 when it is invoked from the emc script, it is passed '--get-config', and that causes it to go to the screen that has only cancel/run as options 18:10:15 cancel in that case should exit the program and cause the emc script to exit 18:10:19 yes - one "main menu" 18:10:19 instead it moves to the main menu 18:10:35 in the --get-config case, the "edit" "backup 18:10:48 it's confusing that you have to cancel first, before being able to mke a new config 18:10:49 "restore", "new", etc options should not exist 18:11:01 (and then you need to cancel again, before you can run your new config) 18:11:44 hmmm. maybe I should reconsider my last statement 18:12:24 I was thinking if you invoke the emc script (and thus the --get-config behavior of setupconfig) that running an existing config would be the only option 18:12:31 (to avoid confusion) 18:12:53 if you want a new config, you would invoke setupconfig by itself without --get-config 18:13:01 ok 18:13:07 but newbies won't even know setupconfig exists, so.... 18:13:17 and probably we need to add setupconfig to the menu aswell 18:13:22 isn't it supposed to make a default config, so I don't get the menu the second time? 18:13:24 emc2 (for running) 18:13:31 emc2-config (for chosing) 18:13:57 SWPadnos: we never really decided about defaults 18:14:12 can we please please please put the run script in the emc2 dir instead of the scripts dir? 18:14:34 so the naive user can find it? 18:14:38 SWPadnos: why? 18:14:51 ok - I thought the original idea was to have you choose, then it would use that setup by default from then on 18:15:09 that was discussed 18:15:16 because it's not obvious where to look for "emc" - there's a scripts dri, a bin dir, a tcl dir (with its own scripts and bin dirs) ... 18:15:22 right 18:15:28 SWPadnos: after compile you get a message 18:15:31 cradek suggested a "use this configuration next time" checkbutton on the run page. 18:15:33 and the README is prety clear 18:15:39 and if I install a .deb? 18:15:45 jepler: that sounds like a nice idea 18:15:49 SWPadnos: cradek's .debs have /usr/bin/emc 18:16:00 SWPadnos: no, after you install a deb you have to type 'emc' 18:16:06 or use the menu 18:16:07 there's no executable in the emc2 dir, so it's not obvious wher eto go 18:16:11 ok - that's fine 18:16:18 there is no emc2 dir for installed deb's 18:16:25 it's not "discoverable" 18:16:26 too many conversations at once 18:16:28 heh 18:16:32 configs are in /etc/ , run script in /usr/bin, etc 18:16:49 installed, there should be no problem, everything is in the right place 18:16:49 ok, back to setupconfig.tcl 18:16:58 run-in-place, read the README 18:16:58 why should it be in the scripts dir though? is there a good reason (not "because it's there") 18:17:02 run-in-place is for developers 18:17:08 SWPadnos: just that 18:17:18 and because you'd have to change some things to move it 18:17:30 ok 18:17:38 and because I see no advantage in having it in emc2/ 18:17:46 minimizing clutter in the top level dir... but arguably the main executable isn't clutter 18:17:58 it then becomes painfully obvious how to run the program 18:18:17 jmkasunich: remember you'll have to have 2 18:18:18 can we go back to setupconfig.tcl? 18:18:19 emc.in and emc 18:18:19 SWPadnos: "type the name of the executable, no matter where you are in the filesystem"? That's the unix way. 18:18:29 afaict, cradek's deb delivers on that 18:18:38 jepler: installed systems work ok 18:18:43 sure - the .deb install would be OK - that's good 18:18:51 can we go back to setupconfig.tcl? 18:18:52 ;-) 18:18:57 SWP & I will go to emc-devel for further talk on run-in-place 18:18:58 I really don't want to talk about setupconfig.tcl 18:19:10 darn 18:19:11 a guy puts in a jpeg and suddenly everybody's asking him to bless their babies 18:19:14 anyway - I't just a thought. people trying it out, or trying a new version (without installing over their old, working version) might have an easier time 18:19:43 jepler: like it or not, you know more about usability than I do 18:20:15 even I think there is something wonky about the top levels of menus tho 18:20:30 it sounded good when I wrote it, but when I use it, wonky 18:20:35 jepler: we're not asking for you to change the code.. 18:20:50 we're asking for your not so humble oppinion on it ;) 18:21:38 I agree, there's something fishy about it 18:22:00 ok, I think basicly we have 2 tools in one piece of code 18:22:00 so I'm trying to brainstorm how to make it better 18:22:12 1. configchooser, 2. config-helper 18:22:31 yes - and that first menu is vasically a tool-chooser 18:22:44 basically even 18:23:24 I think the part for choosing the configuration works great 18:23:29 how about having these buttons on the main screen: "Quit", "Advanced", "Run Selected" (or "Start EMC") 18:23:46 SWPadnos: main screen of what? 18:23:54 setupconfig.tcl 18:24:09 jepler: Can you bless these 45,000 babies on Saturday? 18:24:13 there are three "main screens" (kinda) 18:24:19 I think the main wonkiness comes from there being only two buttons on the main screen - run and quit 18:24:19 If creating a new configuration is what every user needs to do the first time they run setupconfig, you can't call it 'Advanced' 18:24:43 you don't have to create a new config - you can use "sim" for example 18:24:44 I think I kinda liked the initial idea 18:24:54 1. have the users create a new config the first time 18:25:00 the one that says "config/quit/run", the one that says "pick a config and hit run" and the one that says "pick a config and hit one of edit/backup/restore/delete/new/quit" 18:25:13 2. start with that config already selected, and only one button away from running emc2 18:25:23 2.a. but also allow to go back and create a new config 18:25:59 how's that as a plan? 18:26:22 what if we had only one "main screen" that says "pick a config and click run/edit/backup/restore/delete/new/quit" 18:26:36 run/edit/backup/restore/delete/new/quit is a *lot* of buttons 18:26:36 if called with a config name, that config would be pre-selected 18:26:48 the very first time tell the user a bit about emc (where to get help & all), and assist him in setting up the config? 18:26:54 that would work, but hiding the less frequently used functions would be better 18:27:18 most frequently used: run. least frequently used: delete? new? 18:27:22 after that is taken care, it would start up with his config already selected, and 2 buttons (RUN & OTHER) 18:27:32 jepler: quit ;) 18:27:41 quit should be there, for sure ;) 18:28:12 quit will be infrequently used tho, once emc starts, when you exit emc itself you don't go back to the menu 18:28:22 why not let the user use the file browser to backup, restore, and delete configs? 18:28:38 jepler: because it involves files from various folders 18:28:44 what file browser? 18:28:50 a naive user doesn't know that a "config" is a directory 18:28:52 SWPadnos: whichever one 18:28:56 it also rewrites parts of the files 18:29:09 OK, maybe I don't understand all that backup/restore do 18:29:15 (or was that taken out?) 18:29:21 time out 18:29:26 jepler: backup/restore takes a folder and tgz'es it 18:29:31 right 18:29:54 copy actually copies stuff from common/ into the new config dir, and rewrites the ini to use it from there instead of ../common/ 18:29:56 new is the only one that does really weird stuff - gets files from multiple directories, modifies the ini file to change paths, etc 18:30:03 what he said 18:30:25 so maybe save/restore is really not that necessary 18:30:36 although it can help users 18:30:42 it would be helpful for support 18:30:44 backup/restore/delete could be done with a file browser or cmd line, if the user understands that a config is a dir 18:31:04 backup and restore are very much intended for support 18:31:12 you backup and email that to someone, they fix it and mail back to you, then you restore the fixed config 18:31:21 I've had enough of asking folks to send their config info and getting only part of it 18:31:41 jmkasunich: yeah, but I often asked people to send their ini file, and got only part of it.. 18:31:47 heh 18:31:48 so not sure if this will work either 18:32:11 ok, back to the big picture 18:32:11 for some folks, nothing will work 18:32:21 Thousands of times I've gotten only the unessential part of a config. 18:32:22 I still think we need 2 distinctive things 18:32:29 but if you tell them "click on backup, and mail me the file it generates" 18:32:30 one for running (choosing a config) 18:32:45 one for admin stuff (new config, backup/restore, etc..) 18:32:58 alex_joni: that's the division that exists now 18:33:02 they can be one file.. 18:33:14 except that the first time you try to run, you probably actually want to do a "new" first 18:33:15 but it's clunky now - having to cancel to get to the config management is weird 18:33:17 jepler: right, but the first one is always run 18:33:34 then having to cancel again to get a run option is even weirder 18:33:46 if I were a user I'D want this: 1. apt-get install emc2 18:33:51 2. run it for the first time 18:34:07 3. some screens appear with basic information, and where to get more info, support, etc 18:34:08 SWPadnos: So instead of 'Cancel', have 'Maint' and 'Quit' buttons (at the run page) 18:34:30 right - "Quit", "Advanced / Maint", "Run" 18:34:35 4. a screen that allows me to make a new config (after I've learned that's the way I should do this) 18:34:38 or whatever word we can come up with for 'new, backup, restore, delete' 18:34:52 "manage Configs" 18:34:55 SWPadnos: you can't call it "advanced" if every user has to go into that section on the first run 18:34:59 5. after I've done that, that's remembered as a default config 18:35:13 6. a run-page appears with my default config selected 18:35:15 as alex is saying, it should go there the first time by default 18:35:32 7. any further runs will start at 6, but allow to go back to 4 18:35:54 also, there should probably be a "Help" button somewhere ;-) 18:36:05 now that I think of it, the run-page should also have a make default button 18:36:18 in case I want to change the default without going back to 4 18:37:09 or, have several menu items "Start EMC" (starts with last configuration or setupconfig if no last configuration is known), "Start EMC (safe mode)" (starts setupconfig with config chooser and run button), "Configure EMC" (starts with new/delete/backup/restore) 18:37:15 a brand new user doesn't neccessarily want to make a new config as the first thing they do 18:37:21 they might want to run the sim config 18:37:25 It's true, not a user of a turnkey system either 18:38:06 jepler: I like that 18:38:10 in the turnkey case, the default should be preset, so it would go straight to the run screen with that config pre-selected 18:38:46 maybe : Start EMC (default config), Start EMC (choose config), Configure EMC 18:39:18 I like alex's suggestion 18:39:25 the real world calls. you guys can finish hashing this out 18:39:42 if no default is present, the first choice would be greyed out 18:40:05 when I said "menu", I meant something like the gnome Application menu 18:40:08 not a part of setupconfig 18:40:08 jmkasunich: I was talking about 3 gnome menu items 18:40:11 and jepler too 18:40:19 or maybe KDE menu 18:40:33 huh? 18:40:49 then everything you are talking about applies only to installed versions 18:40:50 point & click 18:40:57 jmkasunich: mostly yes 18:41:05 bah humbug 18:41:41 we need to worry about installed versions, primarily 18:42:01 users that can run CVS stuff, can probably figure out the current layout 18:42:22 but the current layout is not friendly enough for installed systems 18:42:23 I know, but how are those buttons gonna actually do their thing? by invoking either the run script or setupconfig, and passing a param 18:42:30 yes 18:42:42 so you can do that from the command prompt for run-in-place too 18:42:44 so the actual behavior is still driven by setupconfig 18:42:49 right 18:43:21 so setupconfig has to be able to do the things we want anyway 18:43:52 right 18:44:01 but it mostly does them right now 18:44:37 if they click the GNOME button Config EMC, it will invoke setupconfig, and they can make a new config. IMHO, they should NOT have to exit setupconfig and click another GNOME button "Run EMC" to run the new config 18:44:48 they should be able to do that within setupconfig 18:44:52 right 18:45:28 in that case, we're back to the three "main menus" in setupconfig. and the talk of GNOME buttons is irrlevant 18:45:43 incidentally, this brings up an interesting problem I see when running emc2 from a KDE desktop icon 18:46:01 I agree that we should support three gnome buttons - its behavior should depend on the args it is passed 18:46:14 the bouncy icon "starting program" indicator stays there for like 10-15 seconds, even after the UI comes up 18:46:27 so? 18:46:46 or if no args, then the defaults found / not found in a file 18:46:56 alex_joni, was that for me? 18:47:00 SWPadnos: yes 18:47:09 oh - well, it doesn't look right ;) 18:47:14 lol\ 18:47:34 I have the axis display up, loading 3d_chips, and the cursor still has a bouncing icon 18:47:47 SWPadnos: complain to KDE people 18:47:55 and send them my regards for such a shitty job 18:48:00 how about when you are using mini, and the splash screen disappears, it looks like the startup failed, then 10 seconds later mini finally explodes onto the screen (fscking huge that "mini" is) 18:48:03 I can do that, but it doesn't happen with all desktop icons 18:48:22 especially when spread across 3840x1024 triple screens ;) 18:48:44 hardly mini in that case 18:49:20 I wonder how the bouncy kde icon knows when to stop bouncing? 18:49:44 I don't know - I just noticed it, and figure I'd mention it to you guys, so it's in the back of more than just my mind 18:49:55 thanks a bunch 18:50:06 heh - not like it's cluttered or anything ;) 18:50:09 First thing I configure to off 18:51:02 the splash screen is a similar issue - it should pop up asap, and say up until the GUI appears. Staying after the GUI appears is annoying, disappearing before the GUI appears is worse. 18:51:17 right now it's a fixed time period, with no conenction to anything else 18:51:42 jmkasunich: that might be done with cradek's idea 18:51:53 jmkasunich: not running emc components with & at the end 18:52:09 jmkasunich: but wait for them to fork and end when init finished 18:52:10 how else can you run them? 18:52:24 that way just before the interface you can kill the splash 18:52:54 or maybe just add a killall popimage 18:52:59 before the interface 18:53:06 and increase the splash time to 20 secs or so 18:53:14 before the interface doesn't help the mini problem 18:53:20 the UI doesn't come up immediately when it's run though 18:53:22 then after ;) 18:53:35 after it returned.. if it would 18:53:36 the GUI's don't have an "after" 18:53:48 it might.. given the same idea above 18:54:02 they return when you kill them, and that begins the shutdown/cleanup of emc 18:54:21 oh, right ;) 18:54:24 slipped my mind 18:54:28 lol 18:54:48 then maybe they should send a SIGTERM to popimage 18:54:55 anyway, we've digressed a lot 18:54:55 but that's nasty 18:54:58 right 18:55:05 like always 18:55:32 I dunno why mini takes so much longer, maybe to solve that problem we look inside mini 18:55:44 no, it's building a list for the backplot 18:55:53 oh 18:55:53 * alex_joni doesn't want to look there.. 18:56:08 waitaminit, list of what? 18:56:17 * alex_joni hides, and points to rayh 18:56:24 you haven't loaded any g-code yet, what is there to build? 18:56:48 err - a big list of null points? 18:57:21 SWPadnos: maybe assigning space? 18:57:28 tcl_very_slow_alloc() 18:57:37 I have no idea, really 18:57:50 tcl_very_slow_alloc_by_name() 18:58:13 maybe the UI programs should just pop up the splash image themselves 18:58:32 that would also help if anyone wants to use e.g. keystick, without X 18:58:56 well.. then there's no reason to use the splash at all.. is it? 18:59:02 exactly 18:59:09 and keystick wouldn't ;) 18:59:09 I thought it's supposed to be displayed till emc loads 18:59:22 unless there's no graphic display available 18:59:23 to give the user a feeling something is happening 18:59:32 emc takes a long time to load, even without mini... the splash lets people know it isn't busted 18:59:33 which is a problem with setupconfig as well 19:00:01 SWPadnos: I see no problem 19:00:14 people setting up no-X emc setups will know how to run 'emc config.ini' 19:00:16 well - you can't run emc2 without X, at this point 19:00:17 heh, a GUI program to select the config, and only once you know the config will you know that it is an X'less config 19:00:23 heh 19:01:17 it looks like keystick is missing from emc2 anyway 19:01:25 yeah 19:01:26 yup, it's not moved in there 19:01:36 so we don't need to worry about that now 19:01:42 ok - then we can ignore that 19:02:06 heh, I found the autosize stuff that makes mini so uge 19:02:09 huge 19:02:30 ok, back to setupconfig people 19:02:32 XSize = min(huge_size, 1000) 19:02:35 ;) 19:02:39 perhaps we need a --size option 19:02:48 s/with/width/ 19:02:49 like --geometry? 19:03:40 wish has that, so it should be pretty easy 19:04:34 I wonder if an "update" command somewhere in mini could draw the main window, before the rest of the time consuming stuff 19:04:54 it would be blank since all the sub-windows and such aren't packed yet, but at least something would appear 19:05:19 how about the popimage, which gets replaced with the real stuff later? ;) 19:05:51 I thikn the splash screen needs to be controlled bu the GUI program 19:05:55 s/bu/by/ 19:06:01 and think, of course 19:06:03 so if no X gui, no splash 19:06:06 right 19:06:13 and also the splash goes away at the right time 19:06:18 autoamgically 19:06:22 garr 19:06:25 not a bad idea 19:06:38 but popimage seems to appear rather quickly 19:06:45 but there is a significant delay while non-GUI stuff is getting started 19:06:46 also simplifies the run script a little 19:06:50 true 19:06:56 right now, popimage is before all that 19:06:59 on my fedora system, the time to get the "%" prompt by typing wish can be 20 seconds from a cold disk. I'm not sure what they did wrong, but it's terrible. 19:07:21 cold disk as in "spun down"? 19:07:35 no, as in "the right stuff isn't in cache" 19:07:37 even on my "primed' machine, it was 3 seconds or more 19:07:41 but I guess I'm exaggerating, it's more like 5 seconds 19:08:00 primed, it's fast on my system: 19:08:01 $ echo destroy . | time wish 19:08:01 0.23user 0.02system 0:00.26elapsed 95%CPU 19:08:18 the "cold cache" case is harder to test 19:09:03 0.71user 0.13system 0:01.27elapsed 66PU (0avgtext+0avgdata 0maxresident)k 19:09:05 0inputs+0outputs (0major+1032minor)pagefaults 0swaps 19:09:48 jepler: I don't think wish is the problem 19:09:58 tkemc uses the same interface, and runs a lot faster 19:10:02 since popimage also uses that 19:10:06 tkemc uses the same wish 19:10:10 popimage is a tcl program 19:10:12 oh, popimage is wish too? 19:10:14 popimage uses wish 19:10:16 yep 19:10:16 I didn't realize 19:10:23 tkemc & mini use emcsh 19:10:25 that's why it takes so long ;) 19:10:28 wish+emc bindings 19:10:40 then it would be easy to "send" a command to popimage, from any Tk-based front end. that covers tkemc, mini, and axis 19:11:07 jepler: if you know how, everything is easy 19:11:07 something like: catch { send Popimage destroy . } 19:11:17 I'll look into it later when I'm booted into ubuntu 19:11:29 are they in a global namespace? 19:13:03 You can find running wish shells from any other wish shell. 19:13:11 ok - cool 19:15:05 heh, it works 19:15:29 (an update in the right place means the mini window pops up right away, and fills in later) 19:23:09 alex_joni_ has joined #emc 19:24:51 rayh is now known as rayh_away 19:26:54 03jmkasunich * 10emc2/tcl/mini.tcl: added some calls to 'update' throughout the initialization of mini, reduces the long delay before the user sees anything. 19:27:27 * jmkasunich has lots of errands to do today 19:27:30 jmkasunich is now known as jmk_away 19:55:19 Imperator_ has joined #emc 19:58:20 Hi all 20:01:05 Hi Martin - how are you? 20:01:40 hey Martin 20:01:45 long time no see 20:14:28 alex_joni_ has quit 20:16:50 * alex_joni goes to bed 20:16:52 night all 20:23:59 SWPadnos is now known as SWP_Away 20:25:35 night Alex 20:26:01 SWP_Away: good, thanks and you 20:29:51 skunkworks has quit 20:47:31 sleep yight alex_joni 20:49:11 hey Jymmm :) 20:49:16 hi 20:50:28 I tried to cook the cat, but he disagree :( 20:51:33 Jacky^ Did you boil it first? 20:51:50 nope .. 20:51:53 :D 20:52:13 try that first.... you can use the tub if you dont have a big enough pot. 20:52:20 haha 20:54:50 http://www.gospelcenterchurch.org/catspill.html 20:57:21 Im having some problem convincing the cat to enter in the pot 20:58:28 Jacky^ 1) Get Gloves 2) Get BandAids 3) Get Cat and toss in pot. 20:58:48 :> 21:00:04 /\-/\ 21:00:05 /a a \ _ 21:00:05 =\ Y =/-~~~~~~-,_____/ ) 21:00:05 '^--' ______/ 21:00:05 \ / 21:00:07 || |---'\ \ 21:00:09 (_(__| ((__| 21:07:37 03jepler * 10emc2/tcl/bin/popimage: 21:07:37 get rid of title bar and extra padding, so that the splash screen is just the image. 21:07:37 compute the window size in a way that doesn't use "after" 21:10:03 skunkworks has joined #emc 21:10:14 staggerlytom has joined #emc 21:10:27 afternoon all 21:11:02 _____ __ __ ____ 21:11:03 | ____|| \/ | / ___| 21:11:03 | _| | |\/| || | 21:11:03 | |___ | | | || |___ 21:11:03 |_____||_| |_| \____| 21:11:07 hi :D 21:11:25 is that macro? 21:11:38 is figlet :) 21:12:01 requires mono spaced font (i didnt) 21:12:12 :( 21:12:24 ;P 21:12:27 fixed now 21:21:31 Hi tom 21:24:29 rayh_away is now known as rayh 21:25:24 03cradek * 10emc2/ (13 files in 8 dirs): first shot at eliminating sudo and being smarter about root privs 21:27:34 Hi Ray, you tried the halvcp? 21:27:55 Nope. Got enough problems of my own making right now. 21:28:01 What you find? 21:28:41 :), it works, the led widget didnt, but i made a text file thatcreated a gui that made real hdwr work 21:29:19 skunkworks has quit 21:29:41 seems that it uses the std gtk widgetset and can have more widgets&combos, uses a packer like tcltk 21:31:40 you link widgets to halsignals and buttons can make linked hdwr work 21:32:29 btw, I got 2 8255 type boards coming in from thailand, 72 io lines each and 64bucks a piece 21:32:41 staggerlytom: pci? 21:32:48 yep 21:32:58 (lookin up the info ) 21:34:01 jepler: if you feel like it, I'd be happy if you would audit module_helper.c 21:34:19 cradek: is it in cvs? if not, where? 21:34:23 yes 21:34:34 I just checked all those changes 21:34:46 emc2/src/module_helper 21:34:53 jepler: http://www.futurlec.com/PCI8255.shtml 21:35:52 jepler: I think aside from module_helper and some ioperm() calls in hal, we don't need root; this will be a nice cleanup. 21:36:25 03rayhenry * 10emc2/tcl/bin/halconfig.tcl: Watch mode cleanup. 21:36:46 roltek has quit 21:38:09 cradek: at first glance, it looks like I can create ~/rtapi.ko with anything in it I like, and have module_helper insert it. 21:38:21 oh, wait, path_whitelist 21:38:22 forget it 21:38:42 k4ts has joined #emc 21:38:59 oh, duh 21:39:09 hello 21:39:18 /whitelist/directory/../../../../home/jepler/rtapi.ko 21:39:18 staggerlytom: 24 signals on 34 pins? 21:39:25 grrr 21:40:35 jepler: is "only one dot in the path" a valid test for that (strchr() == strrchr()) 21:40:43 ? 21:40:56 ray: uses dip headers, pix on website listed above, 3x8255=3xheader & power cnx looks isolated (external anyway ) 21:41:23 ray: a bit too small for me to count in picture 21:41:46 They look like dual row 17 pin headers. 21:42:01 Is there a wiring diagram for it? 21:42:08 ray: yep 34 pins, the zoom lemme count em 21:42:40 yep, a manual someplace, also for the chip 'jet' that is the pci interface to the 8255s 21:43:04 cradek: !strstr(.., "..") ? 21:44:35 ray: www.ett.co.th is manufacture and themanual is in thai :-( 21:46:35 03cradek * 10emc2/src/module_helper/module_helper.c: forbid /whitelist/path/../../ etc 21:47:13 * rayh reaches for my thai-en technical translator; 21:47:32 price is just right. 21:49:03 wow dedicated 82c55 chips. 21:49:26 so the tiger 320 is just a pci connector 21:50:20 rayh: yeah, tiger 320 presents a 4-bit address, 8-bit data bus .. I have been skimming the datasheet 21:50:45 cradek: I can't find any documentation about whether environment variables affect the behavior of insmod 21:51:02 Okay. That sounds like it shouldn't be to difficult to setup a driver for. 21:51:16 HAL of course. 21:51:41 That price even beats a mesa 5i20 21:51:44 jepler: good question. It does get passed along I guess. 21:52:06 cradek: use execve? 21:52:19 I wonder what environment insmod/rmmod need 21:52:22 maybe none 21:53:38 jepler: did you check the malloc and the NULL termination of the argv array in the insert case? 21:53:44 ray: the xilink chip on the mesa is more flixble, if u learn the free ide/sdk 21:54:08 cradek: no; it puzzled me a bit 21:54:20 jepler: then it's probably written badly and should be changed 21:54:33 is mod ever changed from argv[1]? 21:54:39 no 21:54:47 argv[2] iirc 21:54:48 exec_argv = argv + 1; exec_argv[0] = "/sbin/insmod" 21:54:51 er, right, argv[2] 21:55:53 staggerlytom: I really like the loadable ability of the mesa. 21:55:55 I don't know if unix guarantees that main()'s argv is NULL-terminated though 21:56:12 no I bet it's not 21:56:22 obviously, that's what argc is for 21:56:30 ray: your thai-en is ChatChai or the prof who asked for emc help late last year for his students & emc 21:57:10 ray: loadability? is it flash ( wanna move to dev to unclutter for cradex & jepler?) 21:59:34 jepler: seems to work with no env, so I'll commit that 22:00:13 03cradek * 10emc2/src/module_helper/module_helper.c: in case the environment can influence insmod, clear it 22:01:14 cradek: this all assumes that the user cannot control /lib/modules*, /usr/realtime/*, and that the modules themselves are not security risks 22:01:24 of course 22:01:36 there is no way to avoid that assumption 22:02:16 unless I'm missing something... 22:03:01 if the user can control /lib/modules he can already have his way with the system 22:03:18 It was the final assumption I found hardest to believe 22:03:34 true 22:03:46 but that's beyond the scope here I think 22:05:17 night 22:05:29 k4ts has quit 22:06:41 cradek: I'm wonder if something weird happens when the last dot is before the last slash 22:06:55 You really can't protect the emc user's system from her/him. 22:07:42 rayh: when you play with root privs, being as sure as possible your program only does what you intend is wise. 22:08:10 Night 22:08:12 There's always m101. 22:08:14 rayh: also, dropping root privs when you don't need them is wise. 22:08:18 Jacky^ is now known as Jacky^afk 22:08:26 m101 is entirely safe now. 22:08:46 that's an obvious case where "sudo emc" was bad. 22:08:55 Not if you write the referenced file as sudo cd / 22:09:16 I don't understand 22:09:23 $ ./emc_module_helper insert /usr/realtime-2.6.12-magma/./rtapi 22:09:23 module_basename_len = -2 22:09:51 You write m101 as an executable and give it root access. 22:09:59 jepler: you're right, that's bad isn't it 22:10:18 cradek: I'm not sure exactly what it leads to, but .. 22:10:33 rayh: there's a huge difference between a setuid program installed by some deb (ours) and a stupid setuid program written by a user 22:11:10 Yep. we can use as much protection as possible. 22:11:20 as long as that protection doesn't get in the way 22:11:30 like disallowing root login to kde. 22:11:34 aha 22:11:35 $ ./emc_module_helper insert /usr/realtime-2.6.12-magma/.cheese 22:11:35 module_basename_len = 0 22:11:35 insmod: can't read '/usr/realtime-2.6.12-magma/.cheese': No such file or directory 22:11:45 that's certainly not intended 22:12:53 that bug is strncmp(..., ..., 0) right? 22:13:14 right 22:13:22 good spot 22:13:26 if(module_basename_len < 1) error(argc, argv); 22:14:38 03cradek * 10emc2/src/module_helper/module_helper.c: disallow /trusted/path/.module (which bypassed the module whitelist) 22:16:14 logger_aj has joined #emc 22:16:14 topic is: "Welcome to the Enhanced Machine Control forum - Support and development of a linux based CNC control. | Home:www.linuxcnc.org | Regular Developers' meetings every Sunday 14:00-18:00 GMT | wiki up @ http://wiki.linuxcnc.org | EMC usage map: http://www.frappr.com/emctheenhancedmachinecontroller" 22:16:14 Users on #emc: logger_aj roltek staggerlytom Imperator_ Jymmm jmk_away acemi Jacky^afk rayh A-L-P-H-A @ChanServ jepler lerman SWP_Away LawrenceG Ephexis wb9mjn fenn picnet djb_rh bill2or3 CIA-17 ccjoe alex_joni bill20r3 thalx cradek jtr_ robin steves_logging ValarQ websys 22:21:44 insmod: can't read '/usr/realtime-2.6.12-magma/modules/a.ko': No such file or directory 22:21:55 one of the whitelisted modules starts with "a" 22:22:57 so I should strncmp(..., ..., strlen(module_whitelist[i])) instead 22:25:15 03cradek * 10emc2/src/module_helper/module_helper.c: 22:25:15 allowed /trusted/path/a.ko if one of the whitelist modules started with a 22:25:15 (wrong strncmp length was used) 22:25:17 good grief 22:25:25 how many bugs can I write in 100 lines? 22:25:32 this is why all programs suck, isn't it 22:28:43 Imperator_ has quit 22:30:06 cradek: bbl 23:00:00 staggerlytom has quit 23:06:40 acemi has quit 23:21:45 cradek at least it isn't web based then you'ld have to worry about secruity too. 23:23:52 03cradek * 10emc2/ (src/Makefile src/emc/iotask/Makefile debian/rules): I guess io doesn't need root privs; it doesn't call iopl/ioperm anymore 23:24:13 03cradek * 10emc2/src/module_helper/module_helper.c: whitelist all the drivers 23:27:49 03cradek * 10emc2/src/hal/utils/halcmd.c: (log message trimmed) 23:27:50 Drop root privs immediately when started, but reclaim them in order 23:27:50 to add or remove modules. An alternative scheme would be to use 23:27:50 the module helper and not have halcmd be setuid at all. I think 23:27:50 that would be superior but a more invasive change. 23:27:50 I don't understand what hal lock/unlock does, and it was written 23:27:52 to require root, but doesn't actually appear to? This may need 23:28:53 Jymmm is now known as Red70sShow 23:28:53 Red70sShow is now known as Jymmm 23:42:22 paelscrit has joined #emc 23:42:39 hey 23:42:59 roltek has quit 23:43:31 anyone know any cnc irc channels? 23:43:42 #emc 23:44:12 cradek the ppl in there are mean though. 23:45:50 cad cam software sucks 23:45:59 :( 23:49:18 cradek: still here? 23:53:54 yes 23:54:04 saw your comment about hal lock 23:54:14 I thought you might :-) 23:54:38 the premise behind lock is that you can prevent folks from adding new modules, removing modules, changing HAL config, etc 23:54:52 probably something you want to do once the machine is running 23:55:19 HAL lets you do anything anytime, which is a feature IMO, but it also lets you remove the PID loop while the machine is enables, which is not 23:55:21 hence the lock 23:55:51 requiring you to be root to set or release the lock was the simplest way we could think of to protect it 23:55:56 is this meant to prevent accidental changes or sabotage-type changes? 23:56:05 yes, I guess 23:56:14 a user placed lock could prevent the former 23:56:45 there is a gray area in between accident and sabatoge 23:57:14 so there's no technical reason to require root for that part of hal, only an administrative reason 23:57:22 if you are a machine integrator, you sell a machine to a shop, and a bored 3rd shift operators starts messing around with the config, is that accident or sabatoge? ;-) 23:57:37 I can't spell sabotage 23:57:55 agreed, no tech reason 23:58:06 I'm bad at making administrative policies, so I don't know how to handle this. 23:58:15 same here 23:58:22 The technical parts are all done now, until bug reports, I think. 23:58:32 the lock commands were added to halcmd (and the hal core where the locking is actually implemented) 23:58:50 but none of the sample configs use them, and I doubt anybody knows how 23:58:53 a lock, even if it doesn't require root, is still a preventive measure 23:59:07 in order to screw something up, you'd have to do two delibrate steps, starting with unlock 23:59:11 maybe that's all we need 23:59:20 could be 23:59:50 it might prevent accidents still; it won't prevent sabotage but I doubt it ever really did