#linuxcnc-devel Logs

Nov 09 2018

08:06 AM mozmck: I'm getting this error trying to compile linuxcnc and can't figure out why: hal/user_comps/xhc-hb04.cc:31:20: fatal error: libusb.h: No such file or directory
08:08 AM mozmck: Oh, I might have figured it out now - always happens as soon as you ask someone!
01:08 PM mozmck: can hal signals be connected together?
01:09 PM mozmck: net signal1 pin1 pin2 net signal2 pin3 pin4 and then net sig signal1 signal2
01:14 PM andypugh: Np
01:14 PM andypugh: No, I mean
01:14 PM mozmck: Looks like not
01:15 PM mozmck: That would be handy though.
01:15 PM andypugh: But you might be able to alias sig1 to sig2
01:15 PM mozmck: Hmm, I need to read on aliases
01:16 PM mozmck: andypugh: did you see the multispindle branch dgarr made with some fixes?
01:28 PM andypugh: Not yet, I have been busy making things
01:29 PM andypugh: I don’t even really know what version my “production” machines are running.
01:29 PM andypugh: Though I should put them both on Master as that now has everything that I use that wasn’t in before.
01:40 PM mozmck: I should do so myself. I'm still running emc2 2.3 on my router table in the shop - it works - why break it!
01:41 PM andypugh: Err, possibly you have less reason to say that than most.
01:42 PM mozmck: less reason to say why break it? I'm sure you're right.
01:43 PM mozmck: That is a more major upgrade though and would probably need some hardware upgraded and the OS as well.
01:44 PM Tom_L: configs would be slightly different
01:45 PM mozmck: Yeah - I have easier things to upgrade than that for now.
03:26 PM rene_dev_: andypugh do you know why external offsets uses scale and counts instead of float? that doesnt make any sense to me
03:26 PM rene_dev_: its annoying for stuff like thc
03:27 PM rene_dev_: I would prefer machine units
03:29 PM pcw_mesa: I think it was originally conceived for things like jogging while paused
03:30 PM pcw_mesa: but yeah, machine units would be cleaner for general use
03:48 PM pcw_mesa: add a mode bit for integer (easy jog implementation in hal) or floating point (real time motion tweaking)?
03:51 PM rene_dev_: jog while paused should really not be done in hal
03:51 PM rene_dev_: why is integer easier in hal?
03:55 PM rene_dev_: andypugh does touchy use the same mdi as gmoccypy?
03:58 PM rene_dev_: doesnt look like it... what else uses hal_mdihistory.py?
04:00 PM andypugh: Touchy has no MDI history. Which is actually it’s only real problem as I use it
04:01 PM andypugh: rene_dev_: Yes, JWP in HAL is a kludge. There is a better version in Machinekit, I think.
04:02 PM rene_dev_: better jwp or better external offsets?
04:02 PM rene_dev_: I would prefer floats, makes it easier for thc pid, or wobbly turning
04:02 PM rene_dev_: and I think now is the time to change it...
04:03 PM rene_dev_: ah, Im fixing the mdi in gmoccapy.
04:03 PM rene_dev_: the fact hat it doesnt have a history that can be operated by the keyboard annoys me
04:04 PM rene_dev_: andypugh do you maintain touchy? it doesnt start if the gcode folder does not exists
04:05 PM rene_dev_: OSError: [Errno 2] No such file or directory: '/home/rene/linuxcnc/nc_files'
04:06 PM rene_dev_: what tooledit does touchy use? cant find it...
04:26 PM pcw_mesa: motions jog inputs expect "counts" which I think is why the EO branch used integers
04:27 PM pcw_mesa: so just by wiring the jog count source to EO you get jog while paused
05:08 PM andypugh: rene_dev_: Touchy was written by cradek
05:08 PM andypugh: I use it, but don’t consider it mine
05:09 PM andypugh: rene_dev_: I think it just expects you to edit the MDI code that “touch-off” brings up to change the tool table.
05:10 PM andypugh: Yes, I am not 100% happy with EO using ints, but my impression is that it made it more flexible.
05:40 PM rene_dev_: I don’t agree
05:46 PM rene_dev_: you can just convert what you have to float
07:14 PM andypugh: Cool
08:33 PM dgarr: > do you know why external offsets uses scale and counts instead of float? that doesnt make any sense to me
08:34 PM dgarr: Using machine_units for external offsets could lead to hazards particularly if a machine is turned off and then on with an offset value active on a machine_units pin.
08:34 PM dgarr: Suppose you have set a machine_units offset to 10mm, then turn machine off (either on purpose or by other perhaps unexpected means like an estop), then turn on with the offset value unchanged at 10mm.
08:34 PM dgarr: Do you want the machine to jump at the restart to its current position plus another 10mm? If not, when do you move the 10mm as specifed by a machine_units pin? It is a safety issue avoided by the counts/scale interface.
08:35 PM dgarr: As documented, using scale and counts: "The accumulation register is RESET to ZERO at each machine-on." Changes in counts are honored (just like jog wheels etc.) There is no jump at machine-on.
08:35 PM dgarr: And: "... external offsets are DEFINED as ZERO at each startup (machine-on) regardless of the value of the axis.L.eoffset-counts pins. To avoid confusion, it is recommended that all axis.L.eoffset-counts pins are set to zero when the machine is off."
08:35 PM dgarr: NOTE: if the user application does not follow this recommendation, there will still be no jump -- not as easily done if you specify offset as a floating point machine_units value.
08:36 PM dgarr: I wouldn't want to explain to an injured user that they should have followed the recommended procedure and returned all machine_unit offsets to zero before turning the machine on
08:36 PM dgarr: The scale/counts interface supports: 1) scalable, jog-like movements, 2) fixed (floating point) scaling so movement amount is determined by the product of count*scale, or 3) complex movements resulting if both scale and counts are changed dynamically.
08:36 PM dgarr: Ref: http://linuxcnc.org/docs/master/html/motion/external-offsets.html
08:37 PM mozmck_lp: dgarr: makes sense
08:38 PM dgarr: this interface was not designed casually -- i would caution about unexpected consequences of changing it
09:00 PM dgarr: minor s/product of count*scale/accumulated product of count*scale/
09:17 PM dgarr: or better: accumulated product of delta_counts*scale
10:10 PM linuxcnc-build: build #3519 of 1401.rip-wheezy-rtai-i386 is complete: Failure [4failed compile runtests] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/1401.rip-wheezy-rtai-i386/builds/3519 blamelist: Chris Morley <chrisinnanaimo@hotmail.com>
10:33 PM linuxcnc-build: build #5710 of 0000.checkin is complete: Failure [4failed] Build details are at http://buildbot.linuxcnc.org/buildbot/builders/0000.checkin/builds/5710 blamelist: Chris Morley <chrisinnanaimo@hotmail.com>