jepler: thanks for the commit and the git instruction
jepler: here is a patch for hal_input parameters-to-pins that was submitted a while back but never incorporated: http://filebin.ca/nvbxd
dgarr: thanks, committed
EMC: 03jepler 07master * r4f9616d33bfa 10/src/hal/user_comps/hal_input.py: parameters to pins
EMC: 03jepler 07v2_3_branch * r3e7c5ea75690 10/ (debian/changelog nc_files/M101 src/hal/drivers/serport.comp): Use a proper message level
alex_joni: nice stuff you applied yesterday
I've readed about jacobian for kins, It seems that they must be checked before each move?
micges_work: not really clear to me yet ;)
oh ok ;)
alex_joni: re ja3: in theory for lathe should be used identity kins? or both kins( gantry) ?
I would think that it's joints: 0,1
axes: X, Z
so gantry, or generic kins
you can even use genserkins :D
hmm.. I wanted to make a "simple" change to vismach
but I'm a bit lost ;)
I see self.plotdata in redraw(), and I wanted to make it only visible based on a HAL pin (maybe connect it to spindle enable)
alex_joni: I think on sf there is 2.3.2 packages only
but maybe 2.3.4 will be soon so no bother
no, I'm not in a particular hurry to do 2.3.4.
I'll put 2.3.4 up
* alex_joni still waits for SWPadnos's coffee to arrive
first few sips are already being processed
not over here :P
you should be about through with them already :)
well.. since I don't drink any coffee.. maybe not :P
how about tea then?
cradek_ is now known as cradek
jepler: about http://wiki.linuxcnc.org/cgi-bin/emcinfo.pl/emcinfo.pl?action=browse&diff=1&id=Hardware_Requirements
after I did this, installing emc2-dev put it back.
I think I know the fix, checking...
one or both of these should have |...
cradek: did the spindle keep working?
and, the X amp died too, now
what a swirl
cradek: probably libgl1-mesa-dev|libgl1-mesa-swx11-dev ?
yep I agree
libglu1-mesa-dev isn't affected, I think
(I now agree with jmk that git diff giving a pager by default is a bug)
I very much disagree
git config --global core.pager cat
EMC: 03cradek 07master * r53a3f389f5dd 10/debian/control.in: Don't screw up a system that was installed with forced software rendering
cradek: hm, installing emc2-dev did? or do you mean build-dep emc2?
oops, I typed the wrong thing earlier
it was build-dep emc2
it didn't even ask for confirmation!
that was what really surprised me
jepler: any idea how I could change vismach so that the tooltip doesn't always draw?
alex_joni: I'm not really as familiar with that code as you'd like
anyway, here's where a point is added to the plot:
835 if not self.plotdata or point != self.plotdata[-1]:
and here's where it's drawn:
877 for p in self.plotdata:
there's not really a provision for getting the hal object into that function O.redraw ..
that's what I was thinking too..
I noticed how hard that diff to control.in was to read. Anyone want to weigh in on reformatting control.in? http://emergent.unpy.net/files/sandbox/reformat-depends.patch
(no semantic change intended)
does it still work?
I dunno :P
ha, then I'm against it
actually I wonder how it works now; I added python-tkinter as a build-time dependency, but the right package name seems to be python-tk
the format is accepted by dpkg-checkbuilddeps but I didn't actually build a full package
micges: uploaded 2.3.3 to sourceforge
cradek: did you have a lathe?
thanks for noticing
alex_joni: cool, no problem
micges: I have a few lathes, one is running emc
cradek: it is simple XZ config ?
cradek: I wonder how to config lathe in a la3 branch, identity kins or both kins, what do you think ?
what is both kins?
if KINEMATICS_BOTH will still force the user to be aware of joints (stuff like 'home in joint mode then change to world mode') then no
I agree with that
also, a program with "Y" in it should continue to error like it does now
Y and joints aren't that related
(although there's a bug in that: g17g3i1)
how about separate lathekins?
would that be too intrusive?
cradek: what sf number?
something that maps XZ to 01
what's the goal here?
cradek: skip X_Z defines
since lathes currently work fine
alex_joni: I don't understand what that means
micges: no sf bug. I mean the assumption that programs without "Y" in them can't move "Y" is wrong
I don't know whether that's important, nor do I know whether it's related to the question at hand
it's probably not ..
cradek: micges was wondering, and me too, how a "proper" lathe config should look like after the joint/axes cleanup
not sure I know how to tell whether a config is "proper"
follows the design rules
where can I read them?
well.. the idea is to have joints (numbered 0,1,2, ...)
I'm not being difficult - I just haven't kept up with the goals here, so I hesitate to give answers
and axes (that follow the carthesian coordinates)
for a lathe it would be joints: 0,1
and axes: X,Z
the current trivkins mapping doesn't follow that
ok if you want joints consecutive, you want a KINEMATICS_IDENTITY with x=0 z=2
yeah z=1, like I meant to say
that's why I was wondering if having a lathekins is too much trouble
sounds simple to me
or rather having smthg like gantrykins with runtime definition of mappings
so lathekins with KINEMATICS_IDENTITY and x=0 and z = 1?
make it possible for trivkins to see the used axis numbers so that it can do the right thing for any "missing axis letters" trivkins machines
yup, probably gantrykins is the closes
closest to how it should look/work
alex_joni: but it is KINEMATICS_BOTH, Axis will have joints/axes
then it should be KINEMATICS_IDENTITY ;)
well.. not necessarely gantrykins.. but we should have a gentrivkins with KINEMATICS_IDENTITY, where we can assign joints/axes at runtime
have the motors defined in the ini as joint 0,1
and maybe have [TRAJ] KINEMATICS = gentrivkins
then [JOINT_0] AXIS = X
or something like that
maybe AXIS = 0, so it's easier for module parameter parsing
I get the idea
then it's just a SMOP
if gentrivkins knows what axes are defined, it should be able to choose a default sane mapping. [TRAJ]AXES=XZ
[TRAJ]AXES=XZA / JOINTS=021
maps X=0, Z=2, A=1
default will be XYZ=>012.. overrided by [TRAJ] AXES settings
true - there's only one right answer
I think there's even an intercal operator for this (select some bits according to a mask and then squish them together)
if [TRAJ] will be XZ then it will have X=> 0, Y => NULL, Z => 1
yes it's SELECT (~)
why not have them go in the order we all know and love (XYZABCUVW)
cradek: having both definable has an advantage
you can set up things like XYYZZABC
or whatever gantry combinations there are out there
it's an n-to-m mapping
well, yes and no
n-axes, m-joints with m>=n
trouble is, the following error code, and who knows what else, requires joint knowledge outside of HAL
that would be what a lot of people want, assuming we can include solving the homing problems
I thought ferror is a joint problem and is entirely inside hal
err, inside motion, that part of which is in joint space
aren't the offsets (from "machine home") maintained by task/motion/whatever
still in joint space
if jmkasunich shows up later, can one of you ask him to take a quick peak at the change I did the other day?
I'm confident it's fine .. but having him looking over it can't hurt
[19:11:51] <alex_joni> http://git.linuxcnc.org/gitweb?p=emc2.git;a=commitdiff;h=b0e344c2607e53601e3b1e1a073319a3cdbf3037
Author: Alex Joni <juve@juve-desktop.(none)>
Date: Sat Aug 15 02:11:47 2009 +0300
the commitdiff is not that usefull to read
looks like some more of these got in
yeah, I noticed afterwards my git email wasn't set up on that VM
it's too bad git doesn't issue an obvious complaint in that case, instead of just committing
what code between the new location of 'copy joint positions' and the old location uses positions?
the next use seems to be on line 1188 which was already after the copy
I don't think there is any code that should be affected
the only change is that now positions is updated even if we enter the other branch
you mean for the kinematicsInverse call at line 1238?
jas, opening the file now
cradek: what are you doing browsing cnczone? ;)
I wanted badly enough to contact someone there that I figured out my password
and then how could I resist answering a question with a link to documentation?
Discuss Enhanced Machine Controlers here!
surely it's spelled "Controler's"
Results 1 - 10 of about 356,000 for controler's.
good night all
* BJT-Work heads out to put the tower together :)
EMC: 03jepler 07master * r6b0de89db0b8 10/debian/control.in: there is no package python-tkinter
EMC: 03jepler 07master * r9db018cd3f66 10/debian/rules.in: make sure configure exists since it is used to clean
EMC: 03jepler 07master * r7b5358ac7905 10/debian/control.in: wrap very long lines to make small changes more obvious
jepler: a small fix is required for the hal_input.py: http://filebin.ca/xyjjcw
EMC: 03jepler 07master * rf0aebeba25df 10/src/hal/user_comps/hal_input.py: after converting params to pins, must allow pins to be set with setitem
dgarr: keep 'em coming