These are chat archives for machinekit/machinekit

15th
Jun 2016
zhivko
@zhivko
Jun 15 2016 08:56
OK - I builded machinekit on BBB, and it started - stack trace and debugging symbols were included, and I amended issue machinekit/machinekit#968.
I would like to know some more about Machinekit VM for BBB. Is VMWare or some other VM player suggested?
Is it OK to make image with: http://elinux.org/BeagleBone_Black_Extracting_eMMC_contents
and then just play on lets say oracle VM player?
Michael Haberler
@mhaberler
Jun 15 2016 08:57
there is no such thing as a "Machinekit VM for BBB"
you install amd64 debian
this is not an arm-only project
zhivko
@zhivko
Jun 15 2016 08:59
Ah OK. so amd64 debian and build is also building images for BBB ? Does this means I could use Ubuntu 64 to build BBB artifacts ?
Michael Haberler
@mhaberler
Jun 15 2016 08:59
NO
it builds for amd64
how can I possibly explain "machinekit runs on several platforms: amd64, i386, arm7++"
zhivko
@zhivko
Jun 15 2016 09:00
yes - I understand that - i want to setup build platform for BBB to" build from source" for BBB on pc - is this possible ?
Michael Haberler
@mhaberler
Jun 15 2016 09:01
yes, but frankly it does not make sense
zhivko
@zhivko
Jun 15 2016 09:01
I want to speed up build process - look your last comment to me...
I only test on such targets - the build times are ridiculous
Michael Haberler
@mhaberler
Jun 15 2016 09:02
I think you are best served on debugging your problems on an amd64 host, and when done transfer to arm for build
but then if this is a bug in haltalk we will fix it and there is no point
you are missing something - I did NOT recommend to crossbuild
zhivko
@zhivko
Jun 15 2016 09:04
Aaaaah ok now I understand - you suggest to setup debian vm - and do testing there... understood
Michael Haberler
@mhaberler
Jun 15 2016 09:04
yes
developing on an arm target is completely pointless IMO except for hw-dependent stuff which is not the issue here
zhivko
@zhivko
Jun 15 2016 09:05
but for me to find as many possible bugs related to my used architecture - probably testing on BBB is mandatory.
Michael Haberler
@mhaberler
Jun 15 2016 09:05
I gave my advice, it is your choice
IMO the hope this brings you further with a logical problem on the zmq/protobuf level is unfounded
this is not an architecture-dependent bug - you are sending something over zmq which trips haltalk
it is completely irrelevant on which platform you figure this out,, but you will spend > 5x as much time on an arm
zhivko
@zhivko
Jun 15 2016 09:07
You refered as "we would fix it" - I would like to ease this procedure as much as possible - currently I am not in machinekit server stuff and my c development is rather limitted - if you think I can do some more to research my issue please say so - maybe I could workit out maybe not...
Michael Haberler
@mhaberler
Jun 15 2016 09:07
I posted on the issue what you can do
first --> please share which config you are running - I still do not know and I am not standing behind you
second - whatever the config, as I do not know: start haltalk with --debug 1
third: use wireshark to trace the message exchange and post a pcap file
but with some luck step2 will give a hint
zhivko
@zhivko
Jun 15 2016 09:09
Me myself being developer I can understand that just asking when something will be done is VERY anoying but sure if you see something else I can do to make resolution faster please let me know....
Regarding c development - do you you use some framaework let's say eclipse and you could share server eclipse project for development ?
Michael Haberler
@mhaberler
Jun 15 2016 09:10
no, emacs
this is not a development issue though - haltalk trips over some message and we need to know the message
so that eclipse yadayada doesnt help anything with the problem at hand
zhivko
@zhivko
Jun 15 2016 09:11
it is probably more user friendly to do & debug in such (Eclipse CDT) environment than pure C editing? Or how you do it / what do you prefer?
Probably I could simulate that haltalk issue in my Ubuntu / Eclipse / CDT environment ? IS somebody doing devel. in Eclipse?
Michael Haberler
@mhaberler
Jun 15 2016 09:12
eclipse is some editor, it is completely useless for the problem at hand
you do not need to simulate something - please just follow what I outlined above
post your config - or am I asking too much? I do not understand why we cannot sort out the basics and need to discuss eclipse nonsense
zhivko
@zhivko
Jun 15 2016 09:14
So by config - you mean .ini and .hal files ?
Michael Haberler
@mhaberler
Jun 15 2016 09:14
YES
you somehow_ start machinekit, right? so how are you doing the _somehow? nobody knows.
zhivko
@zhivko
Jun 15 2016 09:15
Oh good now I now you need them - I will add them to issue.
Michael Haberler
@mhaberler
Jun 15 2016 09:15
do not copy files there.
zhivko
@zhivko
Jun 15 2016 09:15
yes I am starting from command line:
machinekit ....
pastebin is ok probably ?
Michael Haberler
@mhaberler
Jun 15 2016 09:15
have you read the 'Reporting an error' section on the website
zhivko
@zhivko
Jun 15 2016 09:16
yes
Michael Haberler
@mhaberler
Jun 15 2016 09:16
you likely started from some existing config
which one was it?
if you changed it, commit the changes an push to github
nobody around here wants to read hundreds of lines without an idea what was changed
the diffs are useful
which is why the 'Report an error' section says: push to githubg
zhivko
@zhivko
Jun 15 2016 09:17
yes everything is there on my github
I started from CRAMPS config that came with machinekit
Michael Haberler
@mhaberler
Jun 15 2016 09:18
AHA
and where is the diff
zhivko
@zhivko
Jun 15 2016 09:18
Nothing special changed - just loadusr halttalk in begining - and adding some additional axes... :)
but since my machine is hapilly playing gcode with machinekit (I need to post video of that) - I asume .hal and .ini are OK
I will try to post dif...
Michael Haberler
@mhaberler
Jun 15 2016 09:20
well CRAMPS.hal/ini is from where?
zhivko
@zhivko
Jun 15 2016 09:20
you have this proggy that comes with machinekit that creates configuration for you
yes?
I selected there Arm.Cramps ... and there first version of hal ini came from..
Michael Haberler
@mhaberler
Jun 15 2016 09:21
sorry
very hard to guess "proggy" and "something"
are these copies of some existing configs?
zhivko
@zhivko
Jun 15 2016 09:22
yes!
Michael Haberler
@mhaberler
Jun 15 2016 09:22
great! and which one was it?
zhivko
@zhivko
Jun 15 2016 09:22
they come with machinekit ... wait a sec I will find this name wizard program
Arm Cramps... ughhh will need to turn BBB on to check this...
Michael Haberler
@mhaberler
Jun 15 2016 09:23
bbl
zhivko
@zhivko
Jun 15 2016 09:23
sorry dont keep everything to the letter in my head... I will report back
I will put this info in issue...
Yes :) it is called Configurator selector: http://www.machinekit.io/docs/common/starting-emc/
Michael Haberler
@mhaberler
Jun 15 2016 11:14
when exactly does the failure occur: right after the ping? or during the describe?
ah I see, must be during describe
zhivko
@zhivko
Jun 15 2016 13:14
yes
during describe
Thank you for fast response... I am currently locked out of my BBB this usb connection tethering is pretty lame - sometimes it works sometimes not...
and I cannot use lan at the moment. only usb
I would like to try asap, and trying to connect to bbb, when I suceed and test your patch I will report back
Michael Haberler
@mhaberler
Jun 15 2016 13:16
cany you please try the identity change first as I asked
I need a reference point first which is why I suggested this.
zhivko
@zhivko
Jun 15 2016 13:17
yes ofcourse - but first I need to login to BBB :)
and start machinekit there
zhivko
@zhivko
Jun 15 2016 13:39
I changed identity to: "MyIdentity" and it haltalk crashes also.
zhivko
@zhivko
Jun 15 2016 13:47
I am building now: "Deprecate send container string dest"
Michael Haberler
@mhaberler
Jun 15 2016 13:48
can you post the haltalk log output past your identity change
zhivko
@zhivko
Jun 15 2016 13:49
I am not sure but here is output from machinekit - since haltalk is loaded as normal grogram in .hal probably it should cotains haltalk output also.
Issuing EMC_TRAJ_SET_G5X --      (+224,+144,    +0,    +1,0.000000,0.000000,0.000000,0.000000,0.000000,0.000000,0.000000,0.000000,0.000000,)
Issuing EMC_TRAJ_SET_G92 --      (+227,+144,    +0,0.000000,0.000000,0.000000,0.000000,0.000000,0.000000,0.000000,0.000000,0.000000,)
Issuing EMC_TRAJ_SET_ROTATION --      (+226,+80,    +0,0.000000,)
process error called error True
error service subscribed: True
process status called motion True
status service subscribed: True
process error called text True
error service subscribed: True
process error called display True
error service subscribed: True
sending motion message
process status called motion False
status service subscribed: False
process error called display False
error service subscribed: True
process error called error False
error service subscribed: True
process error called text False
error service subscribed: False
--------------------------------------
[010] MyIdentity
[003] 089402
handle_command_input: req=type: MT_HALRCOMMAND_DESCRIBE

^Cstopping threads
But! I have used loadusr haltalk -d 5
should I change to haltalk -d 1 ?
zhivko
@zhivko
Jun 15 2016 13:56
Do you think I should do "make clean" before issuing make - it will take more time for sure ...
Michael Haberler
@mhaberler
Jun 15 2016 13:57
no, not needed
looks good
and that still crashes haltalk?
zhivko
@zhivko
Jun 15 2016 13:57
? im not there yet
still compiling
? looks good what? :)
Michael Haberler
@mhaberler
Jun 15 2016 13:58
the identity change
zhivko
@zhivko
Jun 15 2016 13:58
ah ok.
I would need to speed up this build process for BBB somehow
Has anybody made crosscompile possible for BBB ?
hmmm.... it is same - it crashess
Michael Haberler
@mhaberler
Jun 15 2016 14:02
as I said: the best is to develop on a PC or a VM, then deploy only
zhivko
@zhivko
Jun 15 2016 14:02
with the new change - I do not see this at console output:
[010] MyIdentity
[003] 089402
handle_command_input: req=type: MT_HALRCOMMAND_DESCRIBE
should I gdb haltalk ?
Michael Haberler
@mhaberler
Jun 15 2016 14:03
let me see
zhivko
@zhivko
Jun 15 2016 14:04
OH
I forgot to tell you
I changed -d 5 to -d 1
so maybe this is the reason not outputing ?
Michael Haberler
@mhaberler
Jun 15 2016 14:05
yes
I think so
sorry, running out of time for the day
cu
zhivko
@zhivko
Jun 15 2016 14:08
(gdb) backtrace
#0  halpr_describe_funct (funct=0xb675eee8, pbfunct=0xe5d00)
    at hal/lib/halpb.cc:96
#1  0x0000fe26 in describe_funct_cb (funct=0xb675eee8, arg=0xbed505e8)
    at machinetalk/haltalk/haltalk_introspect.cc:129
#2  0xb6d8b46c in halpr_foreach_funct (name=0x0, 
    callback=0xfde1 <describe_funct_cb(hal_funct_t*, void*)>, 
    cb_data=0xbed505e8) at hal/lib/hal_iter.c:155
#3  0x0000ff42 in process_describe (self=0xbed505e8, from=0x3b018, 
    socket=0x32f80) at machinetalk/haltalk/haltalk_introspect.cc:44
#4  0x0000eff4 in dispatch_request (socket=0x32f80, from=0x3b018, 
    self=0xbed505e8) at machinetalk/haltalk/haltalk_command.cc:481
#5  handle_command_input (loop=<optimized out>, poller=<optimized out>, 
    arg=0xbed505e8) at machinetalk/haltalk/haltalk_command.cc:60
#6  0xb6bf2e50 in zloop_start () from /usr/lib/arm-linux-gnueabihf/libczmq.so.1
#7  0x0000b980 in mainloop (self=0xbed505e8)
    at machinetalk/haltalk/haltalk_main.cc:113
#8  main (argc=3, argv=<optimized out>)
    at machinetalk/haltalk/haltalk_main.cc:467
zhivko
@zhivko
Jun 15 2016 14:15
ok thx bye!
Michael Haberler
@mhaberler
Jun 15 2016 14:16
can you just run
realtime start
halcmd: loadusr -W haltalk --debug 5
and see what happens
I begin to suspice your config is at work
are you totally sure your config loads without errors?
please:
export DEBUG=5
run your config
post /var/log/linuxcnc.log
zhivko
@zhivko
Jun 15 2016 16:39
I posted it to issue
Also I prepared client to test... I can also test with initial CRAMPS ini and hal?
If you suspect it is related to my config and ini files modification
zhivko
@zhivko
Jun 15 2016 23:07
Is it possible to get diferential update on MT_HALRCOMMAND_DESCRIBE