I'm trying to set up a compile with 8.04 and get the following question during emc configuration.
Is the kernel headers package installed ?
does it say something like "xxx not found. is the kernel headers package installed?"
Yep version.h not found
ok. then I bet you don't have the kjernel headers installed ;)
I presume that these are patched with rtai?
I don't know ifalex/jeff have made a depo that has correct dependencies
(can't type when I have stuff in front of the keyboard)
Know that feeling.
I'll ask Alex in the morning.
take a look at the debian control file (?) in cvs - that'll give you a close list of what you need to compile
I got a listing from the wiki.
is it correct for 8.04?
and installed all those but it was for the sim on 7.10
I suspect that it doesn't use the kernel headers.
ok. isn't that the packages needed to run EMC, not to compile it
[01:18:51] <SWPadnos> http://cvs.linuxcnc.org/cvs/emc2/debian/control.in?rev=1.24
look at the packages for both emc2 and emc2-dev. some of the names may have changed from 6.06 to 8.04 though
I think build-essential should install headers for you
hmmm. maybe not
I did install build-essential
but didn't see headers.
the package you want is linux-headers-`uname -r`
I'm just thinking that there will be others
sure could be
couldn't find the rtai extended version.
I suspect that I'll have to get em from one of the guys that has done the patching.
maybe the guys have some stuff missing in the repo then
what's your uname -r?
I don't see a repo for emc in the sources.list file
oh, maybe it's as simple as that
let me see if I can figure out what it's supposed to be
How do I use that?
same for deb-src
* rayh says showing his ig somthing or other.
don't thank me until it works!
I still don't have a hardy system
unfortunately (?) all my machines work too well to f with
oh right. I was going to burn a copy of the liveCD / beta so I can F with my laptop next week ;)
Install worked pretty well here except for a missed display driver.
This may mean that the package is missing, has been obsoleted, or
is only available from another source
cradek: you were close
the exact line is: deb http://www.linuxcnc.org/hardy
hardy base emc2.2
the base dir contains the infrastructure packages (kernel, headers, rtai, etc)
rayh: your installed /etc/apt/sources.list doesn't have the above URL? in that case I need to fix the LiveCD so that the proper things get installed there
rayh: after adding the URL above, you want apt-get install linux-headers-2.6.24-16-rtai
did he say whether he installed from your CD or the stock beta?
I figured he started with the Ubuntu CD, not yours. he was trying to install RT and EMC2, after all ;)
but if he didn't install from the Live CD, then he should use the install script found here: http://www.linuxcnc.org/hardy/emc2-install.sh
no, he was trying to compile emc2
oh, that would have been easier ;)
I know he used/installed the LiveCD
and I bet it's just a bug in the CD that it doesn't write the proper things to /etc/apt/sources.list
I did install from the emc-live cd.
I'm sort of being a test dummy.
Well dummy anyway.
Seeing what happens and reacting.
I did add both the binary and src repositories to the sources.list but it didn't find linux-headers for me.
I saw one package in there maintained by jeff. The others were stock from ubuntu.
Looking and trying again I tried to install emc2-dev and got this.
The following packages have unmet dependencies:
emc2-dev: Depends: linux-headers-2.6.24-16-rtai but it is not installable
E: Broken packages
rayh: yup, you probably didn't add the proper thing to /etc/apt/sources.list
the exact line is: deb http://www.linuxcnc.org/hardy
hardy base emc2.2
(you probably are missing the base)
where the linux-headers-* is located
Is the same true for the source packages
then sudo apt-get update
and sudo apt-get build-dep emc2
lyx-qt is not available
but the lyx folk have changed the name to just lyx
yeah, I'm aware of that
so it doesn't install anything?
I can install the rest of the stuff a piece at a time.
This will get me there.
I added the info to the livecd.errata
ah. rtai headers are on the way.
[12:14:08] <alex_joni> http://www.linuxcnc.org/hardy/livecd.errata
if there's anything else you notice.. bug me ;)
You bet. That is one of my best acts these days.
I'll do most anything to avoid compiling a real time kernel.
it's sure appreciated
As long as it helps get the release out, I'm for it.
I'm seeing quite a list of languages.
it's a bit more restricted for emc2 :)
english + 10 for AXIS
Ah I see.
[12:19:38] <alex_joni> http://axis.unpy.net/translations
jepler: any ideas how we should handle the packaging problems for 2.2.5/hardy?
alex_joni: did you figure out how to add our repo to the installed sources.list?
I recall you have to monkey with a script the installer runs
alex_joni: there's a facility to modify the build-depends line per os version .. I guess we need to move lyx / lyx-qt to that spot.
cradek: it's in /usr/lib/ubiquity/apt-setup/ apparently
that sounds familiar
jepler: I was thinking asking some debian litarates if it's not doable otherwise (lyx | lyx-qt)
EMC: 03jepler 07v2_2_branch * 10emc2/debian/changelog: From TRUNK: fix #1937281: last move discarded if it's a G1 and the file ends with %
EMC: 03jepler 07v2_2_branch * 10emc2/src/emc/canterp/canterp.cc: From TRUNK: fix #1937281: last move discarded if it's a G1 and the file ends with %
EMC: 03jepler 07v2_2_branch * 10emc2/src/emc/rs274ngc/ (gcodemodule.cc interp_read.cc): From TRUNK: fix #1937281: last move discarded if it's a G1 and the file ends with %
EMC: 03jepler 07v2_2_branch * 10emc2/src/emc/sai/saicanon.cc: From TRUNK: fix #1937281: last move discarded if it's a G1 and the file ends with %
EMC: 03jepler 07v2_2_branch * 10emc2/src/emc/task/emccanon.cc: From TRUNK: fix #1937281: last move discarded if it's a G1 and the file ends with %
alex_joni: that would be even better
oops, I suppose that means I did forget to backport that fix
sure -- I'm glad you mentioned it on friday
jepler: do you remember why we have lyx-qt in there?
there seems to be a lyx package from dapper till hardy
Yea! I've got a running emc2 from trunk on 8.04
Axis does not display the path quite right but does run and display the actual tool path.
The lyx-qt was a left over from the days when there were two graphical tool kits available.
I believe the second tool kit was wx or some such.
lyx-5 uses the qt but doesn't have it in the name.
jepler: it seems we need to treat it separately
16:32 < elmo> sbuild will chose the first of an or'ed dependency
16:33 < elmo> if you're targetting the package at hardy, use the hardy name
rayh: if you think you are seeing an axis bug, can you put up a screenshot or something?
You bet I can. I'll run cds again and capture.
Should I post the text errors here. About 10 lines
rayh: were you able to run cds from TRUNK?
pastebin for 10 lines
which config did you use?
[13:56:37] <rayh> http://www.pastebin.ca/991900
on my trunk (sim compiled on dapper) I don't see any problem loading/running cds
rayh: I get the same error on TRUNK
it's because of the G43 line
do you have a matching tool in your tool table?
k I'll post the screenshot in a bit. What happens is it does not clear the display area and draw the expected path.
ok I bet that's a result of this error
it is failing to run the program to generate the preview
rayh: it fails to load cds and keeps the old preview
running will run with the proper file though
There is a tool 1 in the tbl file but it is zero length.
If I go ahead and run cds, it will display the actual path.
hmm.. might be something
rayh: yeah, it's only a loading file issue for AXIS
are you guys using python2.5? that would be a difference
Yes to 2.5
cradek: yup to 2.5
we're moving with the flow :D
maybe that will mean something to jepler
does removing the G43 line allow the preview to be generated?
Let me try that.
OK that narrows down what I have to look at
and the running path exactly matches the previewed.
does changing it to g43.1 k0.1 give the error?
I'll try that.
I get the error and it does not remove the old image but will plot actual path.
so it's the same as the original problem.
ok so it's not a tool table problem
glad that I can help.
Been a long time since I've had an up to date system.
you're ahead of most of us now...
I don't think that a post of the screenshot will help any.
any neat surprises in store for us at fest this year?
I'll be able to debug this better once I'm at home .. I do have a system with hardy, I just can't access it easily right now
I doubt there will be any surprises.
I agree, the screen shot won't help.
I'm thinking of setting up a teaching lab for fest.
5-6 boxes with 8.04 and a net to a sherline or smithy mill and lathe.
your gcode class seemed to hold a lot of people's interest a couple years ago. maybe it would be cool if they could write a simple program and see it run.
If I can get a projector we can study common problems at the front of the classroom using vnc.
And we can all work up presentations and projects.
I got the go ahead from maxine to buy the boxes.
I believe that she can also supply the projector.
cradek, What is the g43.1 and k param?
0.1 inch tool length offset for a "transient tool"
[14:14:18] <cradek> http://linuxcnc.org/docs/2.2/html/gcode_main.html#sub:G43,-G49:-Tool
I added this to make tool length probing easy (for collet machines etc.)
[14:16:13] <cradek> http://cvs.linuxcnc.org/cvs/emc2/nc_files/tool-length-probe.ngc?rev=1.4
^^ here's how I do that. I bet a lot of sherline folks would like that setup too.
looks good to me.
Is the axis plot from commanded or actual position.
I have not made a good tool length probe yet but I hope to one of these days
it is commanded
rayh: I think you can select from the ini, by default it's commanded
no, the plot is always commanded
^ -> shows how much I know :(
This stuff gets pretty complex in a hurry.
EMC: 03jepler 07TRUNK * 10emc2/src/emc/rs274ngc/gcodemodule.cc: Fix an error seen with python2.5: interp_error > 0 but no Python exception set
I think so
let me try it.
if you can that would be great
works with the g43.1 and kxx
and I do see it pick up the tool length offset at that line.
also works with g43 h1
good -- can we call this one solved?
another one got squashed today ;)
I guess this one won't affect emc2.2 as it's related to TLO_IS_ALONG_W (e.g., 5axis stuff)
I guess that's i.e., not e.g.,
Seems to mess up if I have a tool length in the tbl file.
nope skip that.
let me test a bit more.
I'll read back on the IRC later to see what you conclude
sorry for the moment of panic.
I can see it pick up the offset just fine and then go to work.
That was a quick bug fix.
rayh: they all are lately ;)
We should be ready for an 8.04 rollout soon after Thursday's release.
well.. I still need to build a new LiveCD..
and we need to package/release 2.2.5 for it
hmm.. I think I'll have to backport the fix on halui for 2.2.x
but I would appreciate if anyone else would test first
(the halui_halvcp has the needed infrastructure for this..)
could be 2.2.5~cvs1 if we aren't quite ready to call it 2.2.5 yet
jepler: works for me
did you submit the fix for pyvcp, or should I?
if you have it handy, please do ..
I don't have my hardy machine handy
it's in #emc-board :)
ok, I'll test on dapper, and commit if it still works
I could test here with either.
it seems unlikely to hurt dapper (or systems in general where this fix isn't required)
at worst, it'll add an unmeasurably small time to start-up when pyvcp is used
hmm.. the patch is too borken by IRC to apply directly, so I'll do it by hand
it involves adding one line .. you just have to figure out where :-P
yeah, I think I have :)
my guess was before the import xml.*
EMC: 03alex_joni 07v2_2_branch * 10emc2/lib/python/vcpparse.py: add a workaround which enables finding needed xml packages for hardy
* alex_joni runs home .. bbl
EMC: 03alex_joni 07v2_2_branch * 10emc2/debian/changelog: note pyvcp fixing for hardy
I don't know what vcp was like before but with halui-halvcp config now I can enter an mdi command and then abort or estop from the vcp panel.
All of the functions like coolant, override, pause, resume all seem to work.
rayh: have you played with halui's mdi? I put it in so I can have a "quill up" [MDI g0 g53 z0] hard button etc.
manual jog buttons do not switch mode so don't work.
the mdi problem was specifically with mdi commands started through halui, not ones issued in the regular GUI
I saw the quill up but didn't try it.
what speed is the quill up, g0?
we really need to expand z motion in these configs.
you could make the button do G1 F- instead of G0. It's purely up to you
yep abort and the rest work with quill up
Nice job guys.
quill-up is not special, it's just one use of halui's MDI that I think would be useful
You bet. I just used it to test other buttons during the execution of an mdi command from vcp.
that's an important bug to have fixed .. thanks Alex
argh, another axis keyboard jogging problem
What's it not do?
select X axis. Hold down = to jog towards +X. Press Y. Release =.
(X keeps moving)
yea it does.
I hate keyboard as a machine tool interface.
I guarantee this one isn't as easy to solve as the bug I fixed earlier
you could abort any current jogs when switching axes but I bet that will break simultaneous jogs
how about abort them if they came from -/=?
that's a solution I had never considered; it may be doable
hm I don't seem to know whether the jog was started by = or another method
catch you all later.
jepler: I think the proper fix is having a lookup for active jogs
and operate on that
although the -/= and x/y is still tricky to fix
I was cooking up something along these lines: http://emergent.unpy.net/files/sandbox/yajog.patch
seems safe enough
EMC: 03alex_joni 07v2_2_branch * 10emc2/src/emc/usr_intf/halui.cc: backport fix for #1929461 - halui stops reading inputs during a MDI, also add uvw position info
EMC: 03alex_joni 07v2_2_branch * 10emc2/debian/changelog: note halui fix
cradek: did you hack an existing generator in /usr/lib/ubiquity/apt-setup/generators or did you write a new one?
(wishful thinking that you might remember..)
I made a new one
the file you want to append to is passed in as $1
yeah, I figured
echo "deb ...." >> $1
not sure what 'set -e' does
if any command errors, abort execution and exit the shell with an error
The range of F values is F0 to F20 but EMC seems to treat all values above F15 as the same.
where did you read that?
on their gcode tutiroal
he also says M6 doesn't wait. I think the turotial is just out of date
where is that?\
I can't find it
information / cnc links / introduction to gcode
oh it's even a different site. oops
heh.. "stainless steel could be as low as F0.1"
their CNCUtil library looks interesting
that's modern art :)
not very circular, but interesting
I did a lot of plunges with a small endmill into steel for a project recently. I programmed F0.8 but I think that's the slowest feed I've used. It cut really nice at that speed. I plunge at about 1/3 the sideways feed (2.5 in this case I think).
I've used low feeds on the lathe - if the spindle is only doing 180 RPM and you want 0.001 per rev for a nice finish pass it gets slow