#emc-devel | Logs for 2010-11-19

[00:19:09] <cradek> hm, running my probing programs as mdi o-call macros makes them flaky.
[00:19:53] <cradek> I saw various screwups, including one where the probe was in danger.
[00:22:58] <cradek> but -- often it works fine.
[00:29:42] <SWPadnos> that's not ideal
[00:34:54] <SWPadnos> cradek, do you know if there are any changes on ja3 that make it required for Stuart's robot?
[00:35:34] <SWPadnos> I ask because I foolishly made my changes on master, and I'm about to check in the "unrotate" change on master as well, and I'm wondering if I need to merge those changes into joints_axes3 for Stuart
[00:46:15] <SWPadnos> cradek, do you think it's worthwhile to add pins/params to show when the various matrix inversions don't converge?
[03:32:01] <CIA-2> EMC: 03seb 07master * r691597e0a36a 10/src/libnml/posemath/posemath.h: get rid of an unused structure member
[03:42:42] <cradek> you should be able to easily merge master into ja3
[03:43:09] <cradek> I think it's best (by far) to play with the robot on ja3
[03:43:22] <cradek> someday, ja3 needs to be merged into master
[03:53:49] <SWPadnos> ok. I checked for differences in genserkins.c, and it's only a few printfs (when matrix inversion fails) and the YUCK lines :)
[03:54:02] <SWPadnos> differences between ja3 and master that is
[03:56:46] <cradek> good
[04:02:15] <SWPadnos> it's amazing how much time can be spent looking for one single screw
[05:39:06] <CIA-2> EMC: 03seb 07master * ra94ee9fadb4a 10/src/libnml/posemath/ (posemath.cc posemath.h): remove more dead code
[06:20:27] <CIA-2> EMC: 03seb 07master * ra893e95c1563 10/src/emc/ (motion/usrmotintf.cc motion/usrmotintf.h usr_intf/usrmot.c): dont pass giant structs by value, pass them by reference
[06:29:55] <CIA-2> EMC: 03seb 07master * r5358add559b3 10/src/hal/hal_lib.c: initialize new hal signals to zero more clearly
[06:45:45] <CIA-2> EMC: 03seb 07master * r2cec772c5332 10/src/hal/halmodule.cc: fix broken safety check
[07:46:26] <SWPadnos_> SWPadnos_ is now known as SWPadnos
[10:18:27] <CIA-2> EMC: 03cmorley 07pncconf * r90875e3c19b1 10/ (18 files in 12 dirs): Merge commit 'origin/v2.4_branch' into pncconf
[10:19:45] <CIA-2> EMC: 03cmorley 07pncconf * r0d62d837fdbb 10/src/emc/usr_intf/pncconf/pncconf.glade: Fix progress bar not displaying on top of mesa page
[10:21:30] <CIA-2> EMC: 03cmorley 07pncconf * r17d608cc4907 10/ (9 files in 4 dirs): Merge commit 'origin/master' into pncconf
[10:22:20] <CIA-2> EMC: 03cmorley 07pncconf * r4502a5037577 10/src/emc/usr_intf/pncconf/pncconf.glade: Merge branch 'pncconf' of ssh://cmorley@git.linuxcnc.org/git/emc2 into pncconf
[10:22:53] <CIA-2> EMC: 03cmorley 07pncconf * r1d6b89bcbed0 10/ (341 files in 97 dirs): Merge commit 'origin/master' into pncconf
[10:22:56] <CIA-2> EMC: 03cmorley 07pncconf * rd1774166f399 10/src/ (5 files in 5 dirs): Merge commit 'origin/master' into pncconf
[10:22:57] <CIA-2> EMC: 03cmorley 07pncconf * r83f66537c6c7 10/src/emc/usr_intf/pncconf/ (pncconf.glade pncconf.py): Add widgets for fo,so and userneededmux16, make gtksignals not save
[10:25:36] <CIA-2> EMC: 03cmorley 07pncconf * rf205bf683cdc 10/src/emc/usr_intf/pncconf/pncconf.py: fix more mux16 trouble
[10:25:37] <CIA-2> EMC: 03cmorley 07pncconf * rf63b3838173c 10/src/emc/usr_intf/pncconf/pncconf.py: more HAL code typos for jogging
[10:25:39] <CIA-2> EMC: 03cmorley 07pncconf * r3ce28d2f9145 10/src/emc/usr_intf/pncconf/pncconf.py: fix mpg counter mode. add mpg-fo and mpg-so code.
[10:25:40] <CIA-2> EMC: 03cmorley 07pncconf * r4bd5ddd3ae45 10/configs/common/configurable_options/pyvcp/spindle.xml: add spindle-at-speed LED
[10:25:41] <CIA-2> EMC: 03cmorley 07pncconf * rdfbd2dd98523 10/src/hal/components/mux16.comp: add optional debounce and graycode decoding to mux16
[10:25:47] <CIA-2> EMC: 03cmorley 07pncconf * r49c028c12e29 10/configs/common/configurable_options/pyvcp/spindle.xml: change name of spindle at speed led for clarity
[10:25:48] <CIA-2> EMC: 03cmorley 07pncconf * r995db173ad36 10/src/emc/usr_intf/pncconf/ (pncconf.glade pncconf.py): start of closedloop steppers, change mesa board designation location
[11:02:50] <mhaberler_> mhaberler_ is now known as mhaberler
[16:19:39] <seb_kuzminsky> good mornings
[16:19:50] <jepler> hi seb
[16:25:02] <seb_kuzminsky> brb
[18:37:19] <Dave911> >it's amazing how much time can be spent looking for one single screw .... especially if it is late, dark, unique, and critical ... It is possible these factors create a tear in the space/time continuum ..... ;-)
[18:37:50] <seb_kuzminsky> important small parts are concrete-soluble
[18:38:09] <skunkworks> flashlight shined parrallel with the floor helps.
[18:38:41] <seb_kuzminsky> cradek's solution is to never lift anything important more than 1" off the workbench
[19:01:18] <psha> btw any reasons why there are no python HAL commands for net and newsig?
[19:01:55] <psha> we discussed with cmorley last changes in gladevcp and he asked if it possible to add them as they are needed in pncconf
[19:06:22] <andypugh> Is this my fault?
[19:06:24] <andypugh> [ 167.816457] kernel BUG at /home/moses/Projects/emc2/ubuntu-lucid/mm/slub.c:2969!
[19:06:24] <andypugh> [ 167.816464] invalid opcode: 0000 [#1] SMP
[19:06:24] <andypugh> [ 167.816471] last sysfs file: /sys/devices/hm2_5i23.0/firmware/hm2_5i23.0/loading
[19:07:49] <andypugh> I dont have a /home/moses.......
[19:10:46] <seb_kuzminsky> andypugh: you dont need one
[19:11:04] <seb_kuzminsky> it's just telling you where the file was that mozmck compiled into the realtime kernel you're currently running
[19:11:04] <andypugh> Good.
[19:11:18] <seb_kuzminsky> the problem is in slub.c, which is part of the kernel's memory manager
[19:11:24] <seb_kuzminsky> is this computer otherwise reliable?
[19:11:37] <andypugh> Ah. I wonder what I have done now?
[19:12:02] <seb_kuzminsky> are you running your smartserial driver on this machine?
[19:12:06] <andypugh> I get the bug message every time I try to unload the hostmot2 drivers
[19:12:38] <andypugh> In this case, no. The 5i23 firmware is a non-sserial version.
[19:13:08] <seb_kuzminsky> hmm
[19:13:43] <andypugh> Let me try a firmware that _wasn't_ built by jepler last night with a warning that there was some expermental stuff in it...
[19:14:14] <seb_kuzminsky> i don't think it's the firmware's fault, though i suppose it could be
[19:14:30] <micges> seb_kuzminsky: hi
[19:14:48] <seb_kuzminsky> the firmware loading part of the hm2 driver processes the metadata part at the head of the firmware .BIT file, so maybe it's getting confused about something there
[19:14:50] <seb_kuzminsky> hi micges
[19:15:24] <micges> seb_kuzminsky: what are possibilities to use 5i20 with some spi or other simmilar stuff?
[19:15:43] <andypugh> What stuff do you want to run?
[19:16:01] <seb_kuzminsky> micges: there is currently no good spi support in the hostmot2 driver :-(
[19:16:50] <andypugh> <pokes seb with a stick> Get On With It!
[19:16:55] <seb_kuzminsky> heh
[19:16:57] <andypugh> :-)
[19:17:21] <andypugh> The sserial structure looks a lot like it, you know.
[19:17:38] <andypugh> (Or I think it does)
[19:17:46] <micges> I have external boards with 'rs485/rs522/whatever I want' and wonder in what direction I must go to comunicate them with pc
[19:17:48] <seb_kuzminsky> yeah, that'll be handy to look at
[19:19:04] <seb_kuzminsky> micges: for now i think your best bet is to try to find the spi comp Jon Geo did and see if you can hack that to run
[19:19:06] <micges> and I have one 7I34 that I would like to use with hm2 dricer
[19:19:13] <seb_kuzminsky> or wait a bit and i might get to it
[19:19:28] <micges> driver*
[19:19:29] <seb_kuzminsky> if andy stops hitting me with sticks ;-)
[19:20:20] <micges> mesa 7I34 - Eight Channel RS-422/485 interface
[19:21:27] <micges> Peter did suitable firmware and I wonder about driver
[19:22:10] <jepler> that just means it has differential line drivers and receivers. a specific device with a specific protocol will need a specific driver..
[19:23:41] <andypugh> Is there a standard disclaimer? A chap on #emc has something a little garbled on the Wiki page he just wrote.
[19:24:15] <andypugh> I have a 7i44 hooked up behind me, they are purely electrical not protocol
[19:25:57] <jepler> sure. but I'm talking about the thing you'd hook to it
[19:26:08] <jepler> is it modbus? is it some undocumented serial protocol?
[19:26:36] <andypugh> micges: For a _specific_ application it is reasonably easy to write a .comp for what you want. The hard part is making it general
[19:26:38] <micges> it is modbus
[19:27:43] <andypugh> Given a SPI-aware bitfile for theFPGA you can load the driver with enable_raw and write a comp to stuff the right values into specific hex addresses on the FPGA
[19:27:49] <jepler> yeah, I doubt that's going to be written any time soon
[19:28:12] <jepler> (modbus in our hostmot2 driver)
[19:28:38] <andypugh> Modbus on the serial port is there already isn't it?
[19:28:48] <jepler> andypugh: not on a hostmot2 serial port
[19:29:58] <jepler> you can't "open" a hostmot2 "serial port" through /dev/anything, so our existing serial modbus implementations are useless
[19:30:43] <jepler> you can use the register interface but then you will have the abysmal transfer rates associated with that method
[19:31:26] <jepler> and again you can't use any existing modbus protocol implementation
[19:32:16] <micges> if 7i34 is pin to pin with one 5i20 port, why can't we make firmware like up to 8 pwms/enc (port 2 and 3), up to 8 rs485 (port 4)
[19:32:57] <micges> and improve driver for that
[19:33:12] <jepler> I don't understand that suggestion.
[19:34:57] <jepler> er, maybe I do
[19:35:15] <jepler> I don't know specifically whether the 5i20 has enough FPGA resources for 8 encoders, 8 pwm, and 8 serial
[19:37:15] <jepler> but the fpga firmware end of things is not where the difficulty is
[19:37:22] <jepler> it's the HAL driver for your specific situation
[19:37:28] <andypugh> micges: The issues are about how you tell the Hostmot2 driver what protocol to use.
[19:37:29] <jepler> somebody needs to design and write it
[19:39:37] <andypugh> Choosing one of 15 possible step patterns is easy. One of 47 Hall Sensor codes was bearable, but there is a semi-infinite number of modbus protocols, most of which have not been written yet. So then it becomes a matter of deciding where to draw the boundary. The danger is ending up with someting very flexible that is too hard to use for the general user.
[19:39:53] <jepler> and there's not just modbus
[19:40:42] <andypugh> In fact, there already is a flexible and general-purpose interface to the serial drives on the FPGA. raw_mode ;ets you read and write directly to/from the FPGA registers.
[19:40:58] <psha> sorry for stupid suggestion but if you are in kernel space, why not to expose serial interface?
[19:41:25] <jepler> psha: that's one possible implementation choice, if the goal is to allow a userspace serial program to run the serial comms
[19:41:47] <psha> then problem with different modbus etc implementations will go away
[19:41:57] <psha> since you may reuse existing one
[19:42:27] <jepler> yes, and the problems of 'it's not realtime' have all arrived. so, for instance, you can't be guaranteed to perform a 16-bit ADC read every servo cycle even if the bandwidth is there
[19:42:36] <psha> you definetly will loose in speed/timings but gain flexibility and it will be pretty easy to setup
[19:42:45] <jepler> other users of (broadly speaking) "serial" devices on hostmot2 will want to do things like read an ADC every servo cycle
[19:42:52] <andypugh> Ah, jepler, that reminds me. Do you use the forums at all? Someone there is trying to use your Arduino interface to send step/dir signals to a set of driver boards.....
[19:42:55] <jepler> but in any case, somebody's got to pony up the time and write it!
[19:43:39] <andypugh> I have hinted they might be being folish.
[19:43:49] <psha> jepler: i've got it. i've suspected that there are realtime tasks for serial interface but was not sure
[19:43:50] <jepler> andypugh: no, I don't use the forums. I don't support that arduino firmware in any way. it is inappropriate for position control even if it was working as designed.
[19:44:26] <jepler> a serial ADC is probably not an RS485 device, but I am trying to make a general point
[19:44:47] <psha> if there is rs232/485 device then it's already non-realtime?
[19:44:57] <andypugh> I think the discussion was about generic SPI
[19:44:59] <jepler> bbl, I don't think I'm participating in this discussion in a way that is being helpful.
[19:45:25] <psha> * psha hides in a corner
[19:46:09] <micges> I have realtime boards that have 12 bit ADC on it
[19:46:29] <micges> with rs232/485
[19:46:55] <micges> and my main point is to conenct them to 5i20
[19:47:10] <micges> or in any way to pc
[19:48:59] <andypugh> micges: http://www.pastebin.ca/1996556
[19:49:22] <micges> but I see jepler's point of how difficult that serail driver in hm2 might be
[19:49:40] <cradek> jepler: I just about missed your talk about pluggable interp. that's really cool.
[19:49:56] <andypugh> With the enable_raw mode you can load values directly into the registers (0x5000 onwards). You could do that for a specific application in a simple .comp HAL module
[19:50:45] <micges> andypugh: that's interesting, thanks
[19:51:05] <andypugh> You need a firmware with SPI modules in it though.
[19:51:57] <micges> thanks again
[19:52:14] <micges> week is over, beer time !
[19:52:15] <micges> bbl
[20:48:59] <skunkworks> spindle at speed should be false until the spindle is at speed - right? (no reason it needs to be true when the spindle is off?) Now that I write that - it seems like a silly question.
[20:49:31] <mshaver> Where are the instructions for making .debs from source? I need to monkey around with some things (8i20) and I want to give the fellow testing things a new deb to install, rather than trying to explain run in place.
[20:50:25] <mshaver> my guess is that spindle at speed should be true if the spindle is off and the speed is zero.
[20:50:35] <SWPadnos> mshaver, look in the debian/ dir in your developer checkout
[20:50:49] <mshaver> Thanks!
[20:50:51] <skunkworks> really - that seems odd.
[20:50:55] <SWPadnos> at-speed should be true whenever required by your machine setup ;)
[20:51:14] <cradek> at-speed should be true when your spindle is doing what it's told
[20:51:18] <skunkworks> ok
[20:51:24] <skunkworks> so it can be false all other times.
[20:51:43] <SWPadnos> cradek, is there a delay or other mechanism for preventing false at-speed detection at spindle start?
[20:51:44] <mshaver> Well, if the spindle is off, then you would expect it's speed to be zero. If the speed is what you expect, then... Of course I have no idea how it really works.
[20:51:45] <skunkworks> who is on first.
[20:52:12] <mshaver> what cradek said
[20:52:16] <SWPadnos> (if at-speed = 1 when off or S=0 and stopped, that could cause an issue when starting and waiting for at-speed to become true)
[20:52:17] <skunkworks> mshaver: actually - now that you say it that way - That is what I was wondering.
[20:52:29] <cradek> SWPadnos: I suppose the normal rapid to start-of-cutting position gives you that
[20:52:48] <cradek> SWPadnos: I never thought of it - I have not seen a problem in use
[20:52:57] <SWPadnos> ok, so spindle is turned on and then there is {usually, always} a move before waiting
[20:53:07] <SWPadnos> waiting for at-speed
[20:53:11] <SWPadnos> ok
[20:53:12] <cradek> yeah in normal use I'd say so
[20:53:15] <mshaver> So, you would AND it with halui.spindle-is-on?
[20:53:26] <skunkworks> there is an at-speed pin
[20:53:47] <SWPadnos> you could, to insure that there is an unknown delay between enabling the spindle and feeding back at-speed :)
[20:53:51] <cradek> I think we are having many conversations here
[20:53:56] <skunkworks> heh
[20:54:18] <mshaver> to simplify matters, I will now go build debs!
[20:54:36] <SWPadnos> ah, the confidence of youth
[20:54:41] <SWPadnos> err, I mean good luck
[20:55:04] <mshaver> I'm 49 today. You cannot be speaking of me.
[20:55:16] <SWPadnos> oh. then happy birthday, old-ish man
[20:55:18] <skunkworks> happy birthday!
[20:55:21] <cradek> !!
[20:55:37] <SWPadnos> (I think you're officially old next year :) )
[20:55:44] <mshaver> thank you. I am feeling old, but trying to ignore it
[20:55:56] <SWPadnos> well, it's a thought process
[20:56:04] <SWPadnos> so happy birthday!
[20:56:04] <mshaver> SWPadnos: yes, that will be the hard one to take
[20:56:07] <skunkworks> I then think I have atspeed pin working in my gearshift16.com
[20:56:07] <cradek> you're not old unless your mom had one of those wind-up baby swings
[20:56:09] <skunkworks> comp
[20:57:06] <mshaver> I remember putting my daughter in one of those - she's 25 now...
[20:57:15] <cradek> haha
[20:57:38] <cradek> KLICK KLACK KLICK KLACK
[20:57:47] <cradek> no I don't know why I thought of that
[20:57:49] <mshaver> yep
[20:59:25] <andypugh> mshaver: How's it going with the 8i20?
[21:00:05] <mshaver> well, we applied power and it didn't explode
[21:00:21] <mshaver> I'm now patching & recompiling
[21:00:59] <mshaver> Tomorrow morning we're going to install the patched emc & the firmware Peter provided
[21:01:45] <mshaver> I hope to make pyvcp guis, both of setup and also for testing
[21:02:44] <mshaver> The thing I don't have figured out yet is: Given an encoder input and the ability to command a torque, how do I maintain a set speed? Something tells me there will be a ddt block in there...
[21:03:15] <seb_kuzminsky> mshaver: or a pid
[21:03:24] <andypugh> Encoders have a velocity output, you can PID the current on that.
[21:03:30] <cradek> yeah sounds like pid
[21:03:47] <mshaver> ok, so 'man pid' it is!
[21:04:16] <cradek> what's it mean to command a torque?
[21:04:32] <cradek> seems like you can command and feedback velocity, or command and feedback (?) torque
[21:04:44] <mshaver> ugh, gain... I've been working with stepper systems for long enough to be spoiled
[21:04:54] <cradek> if I put vise grips on your motor shaft, you can't have both the torque and velocity you want
[21:05:22] <skunkworks> I will take a servo loop any day!
[21:05:29] <seb_kuzminsky> god yes
[21:05:37] <mshaver> if I'm commanding a certain vel, and it's going too slow, I up the torque
[21:05:54] <SWPadnos> it's not a torque command, it's torque control to effect a commanded velocity
[21:06:08] <SWPadnos> V in, T out
[21:06:38] <SWPadnos> so the gains will have FF2 or something because torque is more or less proportional to accel
[21:06:57] <SWPadnos> I don't know how P, I, and D are affected
[21:07:03] <SWPadnos> Jon E may know better
[21:07:21] <cradek> oh sounds like you just want a velocity mode pid?
[21:07:24] <mshaver> this is for a spindle, so there's no positioning issue
[21:07:35] <andypugh> Too slow = turn up the wick. Too slow = back it off a bit.
[21:07:37] <cradek> with mesa you get a really nice velocity signal from the encoder
[21:07:46] <SWPadnos> oh, in that case it's better, since you
[21:08:09] <SWPadnos> argh
[21:08:28] <SWPadnos> you're using the derivative of the command for control (like V is the derivative of position)
[21:16:27] <mshaver> So, with respect to building a deb; I looked in the debian directory and nothing jumps out at me, like a README, or an executable file named run-this-to-build-a-deb.sh
[21:17:03] <seb_kuzminsky> mshaver: cd debian; ./configure -a
[21:17:09] <seb_kuzminsky> debian/rules binary
[21:17:26] <seb_kuzminsky> dpkg -i ../*.deb ;-)
[21:17:34] <SWPadnos> yeah. it's really simple once someone who knows how tells you the steps :)
[21:17:50] <SWPadnos> (I almost always have to search the IRC logs to see the right incantations :) )
[21:18:44] <mshaver> jepler has answered this question for me several times, but I couldn't even think of a good search string
[21:18:51] <mshaver> thanks seb
[21:19:28] <seb_kuzminsky> i once saw a tshirt that said "debian/rules" on the front
[21:19:31] <seb_kuzminsky> and on the back:
[21:19:42] <seb_kuzminsky> "the operating system your mom would use, if it were about 5000 times easier"
[21:21:04] <psha> seb_kuzminsky: maybe 'fakeroot debian/rules binary'?
[21:21:34] <seb_kuzminsky> oh yeah, that's better
[21:21:54] <seb_kuzminsky> this has got to be on the wiki somewhere
[21:21:56] <mshaver> i've never done the fakeroot thing myself - don't know what it does
[21:22:05] <seb_kuzminsky> our wiki is getting a bit unwieldy
[21:22:20] <seb_kuzminsky> it's like sudo, but fake ;-)
[21:22:26] <psha> mshaver: it's faking child processes that you are root
[21:23:03] <psha> btw i'd ask another time :)
[21:23:32] <psha> is is suitable to add net and newsig commands to python HAL?
[21:23:39] <mshaver> I'll give this stuff a try...
[21:26:23] <SWPadnos> psha, I remember some discussion about a full HAL interface for python - was it you I was discussing it with?
[21:26:42] <psha> SWPadnos: no
[21:26:51] <SWPadnos> I thought that some of those commands had already been added, but I don't know where or by whim
[21:26:53] <SWPadnos> whom
[21:27:11] <seb_kuzminsky> wham
[21:27:15] <psha> i was talking with cmorley about gladevcp and he asked if it's possible to add theese
[21:28:52] <SWPadnos> the discussion I think I remember is from a couple of months ago or so
[21:29:12] <psha> i know anwser for 'is it possible' question, but don't to 'does it make sense'
[21:29:20] <psha> SWPadnos: i'm on the board only for ~1.5 months :)
[21:29:24] <SWPadnos> heh
[21:30:02] <SWPadnos> during the discussion, we talked about the possibility of making a python-based halcmd, which would add all sorts of features
[21:30:17] <SWPadnos> so python would need all the functions that halcmd needs to do that
[21:30:39] <SWPadnos> I wish I could remember who else was involved in the discussion (and whether anything came of it)
[21:31:27] <andypugh> Hmm, I wonder what is wrong here
[21:31:30] <andypugh> http://www.pastebin.ca/1996627
[21:31:43] <andypugh> I just checked out master and recompiled.
[21:31:46] <psha> is google indexing irc logs?
[21:33:15] <andypugh> It is getting a bit tedious, as I have to reboot every time I exit the Hostmot driver.
[21:33:34] <micges> andypugh: pastebin whole dmesg before emc start
[21:33:43] <andypugh> (The crash is beyond the powers of halrun -U to correct)
[21:35:33] <andypugh> http://www.pastebin.ca/1996635
[21:35:46] <andypugh> Let me reboot and try a non 3pwm firmware
[21:39:43] <psha> SWPadnos: so if adding net/newsig commands is not that bad idea i'll try to find time for this :)
[21:43:08] <seb_kuzminsky> andypugh: i think it's a bug in the tp_pwmgen driver
[21:43:16] <seb_kuzminsky> it calls kfree on a pointer it got from hal_malloc
[21:43:27] <andypugh> Aha! Progress! I have also decided it is a bug in the tp_pwm driver
[21:43:48] <andypugh> It wasn't a problem previously. Hmmm
[21:44:03] <seb_kuzminsky> comment out lines 485-488 of tp_pwmgen.c, recompile & try again
[21:44:17] <seb_kuzminsky> this bug should only bite if you have 1 or more tp_pwmgen instances enabled
[21:44:29] <andypugh> But yes, only a problem with 3pwm firmware and num_3pwmgens > 0
[21:45:30] <seb_kuzminsky> i would guess the tp_pwmgen module is not widely used...
[21:47:00] <andypugh> Indeed not.
[21:47:19] <andypugh> But the problem was not seen on 8.04
[21:47:28] <seb_kuzminsky> hmm
[21:47:33] <seb_kuzminsky> i dont know why that would be
[21:47:40] <andypugh> Which is not to say that it was not a gross and egregious coding fault.
[21:48:30] <andypugh> I can type sudo shutdown -r now in my sleep.
[21:48:34] <seb_kuzminsky> heh
[21:49:24] <seb_kuzminsky> when is hal memory freed? i mean memory from hal_malloc()... is it when the module that called hal_malloc() unloads, or when all of hal unloads?
[21:50:07] <andypugh> OK, deleting 485 to 488 fixes the problem.
[21:50:10] <seb_kuzminsky> yay
[21:50:15] <andypugh> Can somebody push that patch?
[21:50:21] <seb_kuzminsky> i've got a commit for that on the way
[21:50:31] <andypugh> Seems like a no-brainer
[21:51:52] <CIA-2> EMC: 03seb 07master * r176d285a38d5 10/src/hal/drivers/mesa-hostmot2/tp_pwmgen.c: fix a bug unloading hostmot2 when using tp_pwmgen
[21:52:19] <seb_kuzminsky> tp_pwmgen is not in 2.4, so this commit only belongs in master
[22:04:44] <andypugh> Should I now git-pull, check out my dev branch, and rebase?
[22:05:11] <andypugh> Or perhaps I should just check out the dev branch and edit tp_pwmgen?
[22:09:32] <seb_kuzminsky> you should "git fetch" to update your local copy of the central repo
[22:09:54] <seb_kuzminsky> then you can "gitk" to see how your local branch differs from that, and decide from there how to resolve it
[22:10:18] <seb_kuzminsky> do you have a bunch of local commits that you have not pushed anywhere? a bunch of local changes not committed at all?
[22:11:08] <andypugh> A fair few commits not-pushed (nowhere to push to)
[22:12:40] <andypugh> Hmm, can't rebase my branch as there is a rebase in progress. I think git is too clever for me.
[22:16:32] <seb_kuzminsky> you're in the middle of a rebase? you should finish that before doing anything else
[22:18:36] <andypugh> I didn't _know_ was in the middle of a rebase
[22:18:49] <seb_kuzminsky> ah
[22:18:53] <seb_kuzminsky> hm
[22:19:30] <andypugh> I think I shall put the one changed file in a safe place, and delete the branch.
[22:20:36] <seb_kuzminsky> ah yes, the good old nuclear option ;-)
[22:23:18] <andypugh> git gui won't let me delete the branch as it is not fully merged into master.
[22:23:46] <andypugh> I understand the words, just not what they mean when put together.
[22:27:22] <andypugh> Seems to be settled down now
[22:35:29] <seb_kuzminsky> i dont use git gui for things like that
[22:35:46] <seb_kuzminsky> i would move off the branch you want to delete, then "git branch -D branchname"
[22:36:20] <andypugh> I have deleted the branch now. The current problem is reverting to master
[22:36:53] <andypugh> I am in a "detached HEAD state" which does sound plausible ;-)
[22:37:11] <seb_kuzminsky> that's one of the side-effects of using too much git ;-)
[22:37:15] <seb_kuzminsky> try git checkout master
[22:37:50] <andypugh> my branch of master is ahead of origin/master by 8 commits.
[22:37:58] <seb_kuzminsky> great!
[22:38:07] <seb_kuzminsky> did you do the 'git fetch origin'?
[22:38:15] <andypugh> No.
[22:38:22] <seb_kuzminsky> ok do that now
[22:38:27] <andypugh> If that is the magic incantation, I will try it
[22:39:38] <andypugh> And it has brought back one of the files from the branch I deleted. Which is unexpected.
[22:40:33] <seb_kuzminsky> now you can do "gitk master origin/master" and see the commits you've made, and the commits that have been pushed to the central repo by other folks
[22:42:42] <seb_kuzminsky> skunkworks!
[22:43:10] <skunkworks> seb!
[22:44:05] <skunkworks> finally friday!
[22:44:11] <seb_kuzminsky> at long last
[22:44:23] <seb_kuzminsky> man this weekend's going to be great
[22:44:24] <seb_kuzminsky> i can feel it
[22:44:35] <skunkworks> That is awesome.
[22:45:02] <skunkworks> I hope to get further on the k&t tomorrow. after that - I don't know
[22:45:40] <seb_kuzminsky> got any holiday plans for next weekend?
[22:46:40] <skunkworks> no travel - my wifes father is coming home - so we are having a get - together.
[22:47:05] <skunkworks> maybe over christmas vacaation we might travel
[22:47:11] <seb_kuzminsky> where to?
[22:47:12] <skunkworks> how about you?
[22:47:50] <skunkworks> my wife would like to go to amsterdam... but I don't know if that is possible :) might just be a driving trip to somewhere.
[22:47:59] <seb_kuzminsky> i'm taking my wife and kids up to my brother's place in the mountains
[22:48:09] <skunkworks> Nice!
[22:48:21] <seb_kuzminsky> amsterdam is cold & wet this time of year... late spring and summer are nicer
[22:49:04] <seb_kuzminsky> unless you like it cold and wet i guess
[22:49:13] <skunkworks> heh - I will have to tell her.
[22:49:48] <seb_kuzminsky> you might consider timing it so you're there for Queen's Day, that's a fun time
[22:49:55] <seb_kuzminsky> it's in spring some time
[22:50:00] <seb_kuzminsky> big city-wide party
[22:50:02] <skunkworks> wow - you have been?
[22:50:17] <seb_kuzminsky> i lived there for a couple of months back in the 90's
[22:50:26] <seb_kuzminsky> i was back there for work last winter (it was cold & wet)
[22:50:40] <skunkworks> that is neat. I am not that worldly. :(
[22:51:15] <andypugh> It is certainly possible to go to Amsterdam.
[22:51:44] <seb_kuzminsky> where do you live andy?
[22:51:54] <andypugh> Any particular reason for that choice of destination?
[22:52:13] <andypugh> I live about 40 miles from where I can catch a ferry to Amsterdam.
[22:52:33] <skunkworks> my wife missed a chance to go to spain last year - she would have had a day layover in amsterdam. So she is thinking that would be a nice destination. (she got the swine flu the day before)
[22:52:35] <andypugh> But I was talking in more general terms
[22:53:08] <skunkworks> and she checked air prices and they where pretty cheap.
[22:53:32] <skunkworks> honestly - she would travel anywhere pretty much - she just likes to go places ;)
[22:53:42] <andypugh> It's a nice place. There are canals, museums. It's been there longer than most North Americans can get their heads around.
[22:53:51] <seb_kuzminsky> hehe
[22:53:53] <skunkworks> heh -
[22:54:13] <skunkworks> us newbies don't have any history. :)
[22:54:26] <seb_kuzminsky> "this is our historic monument, it's 200 years old!"
[22:54:46] <andypugh> To be honest, the reason most folk from here go there is for the availability of legal cannabis and prostitution, but I think they are rather missing the best bits.
[22:55:19] <seb_kuzminsky> i agree - the hashish is much better than the cannabis ;-)
[22:55:33] <skunkworks> for sure - prostitution is out of the question... ;)
[22:55:58] <andypugh> Don't let that put you off, just be careful what you order if you go into a coffee shop which is oddly short of people drinking actual coffee
[22:56:08] <skunkworks> heh
[22:56:24] <cradek> who would need or want anything stronger than coffee?
[22:56:34] <andypugh> The red light district is just odd. And a bit depressing.
[22:56:55] <skunkworks> well - coffee is the stimulant - then you need the depressant.. and so on ;)
[22:58:29] <skunkworks> oh well - I need to get out in the garage and make a wheel platform for our composter to make it a little more mobile
[22:58:35] <skunkworks> wheeled
[22:58:51] <seb_kuzminsky> mobile composter? weird
[22:58:53] <mshaver> .deb building instructions worked!
[22:58:57] <seb_kuzminsky> hooray!
[22:59:02] <skunkworks> yay
[22:59:12] <mshaver> thanks - now to stick in seb's last file...
[23:00:11] <mshaver> then I'm going to Amsterdam!
[23:00:18] <mshaver> or not
[23:00:22] <seb_kuzminsky> heh
[23:10:12] <andypugh> Far more cost effective if you all send me $10 and I go to Amsterdam for you.
[23:12:55] <mshaver> Ok Andy, I'll paypal you the money. Now, as to the kind of girls I like...
[23:13:05] <cradek> dude, that's like 6 cups of coffee
[23:33:36] <seb_kuzminsky> i'm outta here
[23:33:40] <seb_kuzminsky> see you all later
[23:33:50] <andypugh> Have fun