Re: [Emc-developers] git bisect help

[andy pugh]

On Tue, 14 Apr 2020 at 10:33, Gene Heskett  wrote:

> How, since even the newer, bad code runs thru runtests w/o any problems,

Bisect does not care how you test. But it is probably easiest to run
it on the machine that is showing the problem.

> Can someone suggest a starting branch that ideally would start the bisect
> say at 2 weeks prior to the beginning of the 2.8 merge, just to verify
> it did work then?

2.8 merges in to master rather regularly, but 2 weeks ago would be
7bfebcb23a643c084fc39f1d3bfca9a29c64ae6c
(just from browsing  https://github.com/LinuxCNC/linuxcnc/commits/master)

git checkout 7bfebcb23a643c084fc39f1d3bfca9a29c64ae6c
Will take you back in time 2 weeks.

-- 
atp
"A motorcycle is a bicycle with a pandemonium attachment and is
designed for the especial use of mechanical geniuses, daredevils and
lunatics."
— George Fitch, Atlanta Constitution Newspaper, 1912


___
Emc-developers mailing list
Emc-developers@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-developers

Re: [Emc-developers] git bisect help

[Gene Heskett]

On Tuesday 14 April 2020 06:07:31 andy pugh wrote:

> On Tue, 14 Apr 2020 at 10:33, Gene Heskett  
wrote:
> > How, since even the newer, bad code runs thru runtests w/o any
> > problems,


> Bisect does not care how you test. But it is probably easiest to run
> it on the machine that is showing the problem.

I am. The rest of my machines are running just fine although I have not 
not yet built master on the pi yet which I usually do over the weekends.
>
> > Can someone suggest a starting branch that ideally would start the
> > bisect say at 2 weeks prior to the beginning of the 2.8 merge, just
> > to verify it did work then?
>
> 2.8 merges in to master rather regularly, but 2 weeks ago would be
> 7bfebcb23a643c084fc39f1d3bfca9a29c64ae6c
> (just from browsing 
> https://github.com/LinuxCNC/linuxcnc/commits/master)
>
> git checkout 7bfebcb23a643c084fc39f1d3bfca9a29c64ae6c
> Will take you back in time 2 weeks.

which may be not enough but its building now. Keep in mind I have to make 
debs and install them and reboot to try them as the shutdown is not 
cleaning house as I quit it and I have to reboot to get a clean slate 
for the next try.  That didn't happen when it was working correctly. 
This same machine is one of 2 that are running on a 240 GB SSD, an older 
dell with only 1 sata port. Much faster at about 150MB/sec read speeds.

Got the rpi4b build underway. Ain't ssh wunnerful... :)

2 weeks apparently wasn't enough, its broken there.  How do I back the 
bisect start "good" up another 2 weeks?  That surely should get it in 
good shape.

In the meantime I have to go drop the card for a new mower.  Back later.

Thanks Andy.

Cheers, Gene Heskett
-- 
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
If we desire respect for the law, we must first make the law respectable.
 - Louis D. Brandeis
Genes Web page 


___
Emc-developers mailing list
Emc-developers@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-developers

Re: [Emc-developers] git bisect help

[andy pugh]

On Tue, 14 Apr 2020 at 16:29, Gene Heskett  wrote:

> which may be not enough but its building now. Keep in mind I have to make
> debs and install them and reboot to try them

You should be able to test with a run-in-place. If you are compiling
on the machine under test then you don't need debs.

>  as the shutdown is not
> cleaning house as I quit it and I have to reboot to get a clean slate
> for the next try.

This _could_ be caused by a runtests failure. How does that go?

> 2 weeks apparently wasn't enough, its broken there.  How do I back the
> bisect start "good" up another 2 weeks?

Keep scrolling through "git log" until you find a commit on a date
that suits you.

I still think this is probably a config or hardware problem.
This isn't the command-deriv thing is it? Is that pin netted at all?

-- 
atp
"A motorcycle is a bicycle with a pandemonium attachment and is
designed for the especial use of mechanical geniuses, daredevils and
lunatics."
— George Fitch, Atlanta Constitution Newspaper, 1912


___
Emc-developers mailing list
Emc-developers@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-developers

Re: [Emc-developers] git bisect help

[Gene Heskett]

On Tuesday 14 April 2020 12:35:01 andy pugh wrote:

> On Tue, 14 Apr 2020 at 16:29, Gene Heskett  wrote:
> > which may be not enough but its building now. Keep in mind I have to
> > make debs and install them and reboot to try them
>
> You should be able to test with a run-in-place. If you are compiling
> on the machine under test then you don't need debs.
>
> >  as the shutdown is not
> > cleaning house as I quit it and I have to reboot to get a clean
> > slate for the next try.
>
> This _could_ be caused by a runtests failure. How does that go?
>
Perfectly, no failures.

> > 2 weeks apparently wasn't enough, its broken there.  How do I back
> > the bisect start "good" up another 2 weeks?
>
> Keep scrolling through "git log" until you find a commit on a date
> that suits you.

git log?  What I am doing does not show dates.
>
> I still think this is probably a config or hardware problem.

I can't argue.  Editors have been known to make typu's.

> This isn't the command-deriv thing is it? Is that pin netted at all?

yes, for all 4 axises. I can't stop the following errors without it

And it only effects the y axis. x,z,a work fine.

>From my main hal file:
#
#  AXIS Y = joint_1 *
#
setp   pid_y.Pgain [JOINT_1]P
setp   pid_y.Igain [JOINT_1]I
setp   pid_y.Dgain [JOINT_1]D
setp   pid_y.bias  [JOINT_1]BIAS
setp   pid_y.FF0   [JOINT_1]FF0
setp   pid_y.FF1   [JOINT_1]FF1
setp   pid_y.FF2   [JOINT_1]FF2
setp   pid_y.deadband  [JOINT_1]DEADBAND
setp   pid_y.maxoutput [JOINT_1]MAX_OUTPUT
setp   pid_y.error-previous-target true
setp   pid_y.maxerror .0005


#**
# Step Gen signals/setup for Y axis
#**
setp   hm2_5i25.0.stepgen.01.dirsetup[JOINT_1]DIRSETUP
setp   hm2_5i25.0.stepgen.01.dirhold [JOINT_1]DIRHOLD
setp   hm2_5i25.0.stepgen.01.steplen [JOINT_1]STEPLEN
setp   hm2_5i25.0.stepgen.01.stepspace   [JOINT_1]STEPSPACE
setp   hm2_5i25.0.stepgen.01.position-scale  [JOINT_1]STEP_SCALE
setp   hm2_5i25.0.stepgen.01.step_type0
setp   hm2_5i25.0.stepgen.01.control-type 1
setp   hm2_5i25.0.stepgen.01.maxaccel [JOINT_1]STEPGEN_MAXACCEL
setp   hm2_5i25.0.stepgen.01.maxvel   [JOINT_1]STEPGEN_MAXVEL
#*
# --closed loop stepper signals for joint_1=Y*
#*
# 1st 2 are from motion
net y-pos-cmd<= joint.1.motor-pos-cmd => pid_y.command
net y-vel-cmd<= joint.1.vel-cmd  #  =>  pid_y.command-deriv
net y-output <= pid_y.output => hm2_5i25.0.stepgen.01.velocity-cmd
net y-pos-fb <= hm2_5i25.0.stepgen.01.position-fb => joint.1.motor-pos-fb
net y-enable <= joint.1.amp-enable-out  => hm2_5i25.0.stepgen.01.enable 
pid_y.enable
net y-index-enable  <=> pid_y.index-enable

This is afaik, a copy/paste of x with the x's changed to y's,
and the stepgen incremented to 01.
 
And it all worked prior to the 2.8 merge 2 or 3 weeks back.

Thanks Andy.

Cheers, Gene Heskett
-- 
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
If we desire respect for the law, we must first make the law respectable.
 - Louis D. Brandeis
Genes Web page 


___
Emc-developers mailing list
Emc-developers@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-developers

Re: [Emc-developers] git bisect help

[andy pugh]

On Tue, 14 Apr 2020 at 22:23, Gene Heskett  wrote:

> net y-vel-cmd<= joint.1.vel-cmd  #  =>  pid_y.command-deriv

What's that # doing there?

-- 
atp
"A motorcycle is a bicycle with a pandemonium attachment and is
designed for the especial use of mechanical geniuses, daredevils and
lunatics."
— George Fitch, Atlanta Constitution Newspaper, 1912


___
Emc-developers mailing list
Emc-developers@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-developers

Re: [Emc-developers] git bisect help

[Gene Heskett]

On Tuesday 14 April 2020 17:30:36 andy pugh wrote:

> On Tue, 14 Apr 2020 at 22:23, Gene Heskett  
wrote:
> > net y-vel-cmd<= joint.1.vel-cmd  #  =>  pid_y.command-deriv
>
> What's that # doing there?

IIRC, removeing a later duplicate link.

Cheers, Gene Heskett
-- 
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
If we desire respect for the law, we must first make the law respectable.
 - Louis D. Brandeis
Genes Web page 


___
Emc-developers mailing list
Emc-developers@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-developers

Re: [Emc-developers] git bisect help

[andy pugh]

On Tue, 14 Apr 2020 at 22:54, Gene Heskett  wrote:

> > What's that # doing there?

> IIRC, removeing a later duplicate link.

I don't see any duplication. Have you tried removing it?

-- 
atp
"A motorcycle is a bicycle with a pandemonium attachment and is
designed for the especial use of mechanical geniuses, daredevils and
lunatics."
— George Fitch, Atlanta Constitution Newspaper, 1912


___
Emc-developers mailing list
Emc-developers@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-developers

Re: [Emc-developers] git bisect help

[Gene Heskett]

On Tuesday 14 April 2020 18:27:20 andy pugh wrote:

> On Tue, 14 Apr 2020 at 22:54, Gene Heskett  
wrote:
> > > What's that # doing there?
> >
> > IIRC, removeing a later duplicate link.
>
> I don't see any duplication. Have you tried removing it?


did just now, no diff. It still runs away slowly, a .005 offset, but 
outputs no steps except then the up-down key is being pushed. When one 
of them has been pushed and then released the stepgen output is turned 
off, so the axis is not moving, but the dro marches along slowly as does 
the fb number out of the stepgen until around an inch later it finally 
trip off with a following error. Sometimes the direction of the phantom 
drift is reversed by the opposite key and sometimes the dro continues 
marching along in the same direction until it finally throws a following 
error, but usually moves the dro an inch or better, all of this w/o a 
physical axis movement unless a key is actually pushed.

Cheers, Gene Heskett
-- 
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
If we desire respect for the law, we must first make the law respectable.
 - Louis D. Brandeis
Genes Web page 


___
Emc-developers mailing list
Emc-developers@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-developers

Re: [Emc-developers] git bisect help

[Gene Heskett]

On Tuesday 14 April 2020 19:29:26 Gene Heskett wrote:

> On Tuesday 14 April 2020 18:27:20 andy pugh wrote:
> > On Tue, 14 Apr 2020 at 22:54, Gene Heskett 
>
> wrote:
> > > > What's that # doing there?
> > >
> > > IIRC, removeing a later duplicate link.
> >
> > I don't see any duplication. Have you tried removing it?
>
> did just now, no diff. It still runs away slowly, a .005 offset, but
> outputs no steps except then the up-down key is being pushed. When one
> of them has been pushed and then released the stepgen output is turned
> off, so the axis is not moving, but the dro marches along slowly as
> does the fb number out of the stepgen until around an inch later it
> finally trip off with a following error. Sometimes the direction of
> the phantom drift is reversed by the opposite key and sometimes the
> dro continues marching along in the same direction until it finally
> throws a following error, but usually moves the dro an inch or better,
> all of this w/o a physical axis movement unless a key is actually
> pushed.
>
I seem to have fixed it. I went up the file to the x axis section and 
duplicated a couple dozen lines, changing only the axis/joint 
references, then went down and deleted that same duplicate stanza which 
was now a repeat and surplus.  No clue what I did other than reorder the 
file, but it now works. Go figure.

I did find out one thing though, that I could bring the y table fwd far 
enough to drop off the far side of the home switch. It won't home from 
there as it runs the wrong way looking for the switch, ending that move 
with an error 7.

Now maybe if it quits raining, I can do something about my jungle later 
today.

g'night all.

> Cheers, Gene Heskett


Cheers, Gene Heskett
-- 
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
If we desire respect for the law, we must first make the law respectable.
 - Louis D. Brandeis
Genes Web page 


___
Emc-developers mailing list
Emc-developers@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/emc-developers