Closed brunofavs closed 3 months ago
Will try to fix soon ...
Hi @brunofavs ,
So I tried to fix but will need your help for testing.
Let me know if now you do not have this error.
The fix was this:
because of #900, and for retrocompatibility with old datasets, we will assume that if the transforms field does not exist in the dataset, then the transformation is fixed
The fix should make old datasets be possible to use again. However, the assumption for these is that the transformations are fixed, which is not your case. My feeling is that this will work for everyone else but not for you.
In any case you should try with your old dataset, the script should run but produce bad results.
Correct me if I'm wrong but isn't fixed
and static
the same? Or what is the difference?
Also, in spite of this fix making things work for 99% of prior cases, I reckon it introduces another hidden bug. It's true that it won't work for softbot because it will overwrite all collections with the same transformation, but the same will happen if one was to try to calibrate a transform of the UR10e for instance (hence dynamic). Granted this is highly niche considering the user would usually choose the static transform from the TCP to the sensor or something equivalent.
I will test it with the softbot now. I'm certain it will give awful results despite not crashing.
Correct me if I'm wrong but isn't
fixed
andstatic
the same? Or what is the difference?
should be the same. I am trying to use fixed and abolish the static ...
Ok. I searched both keywords in the entire project and I did find that static
was only really in a few comments not code itself.
Also, in spite of this fix making things work for 99% of prior cases, I reckon it introduces another hidden bug. It's true that it won't work for softbot because it will overwrite all collectionsa
right
but the same will happen if one was to try to calibrate a transform of the UR10e for instance (hence dynamic). Granted this is highly niche considering the user would usually choose the static transform from the TCP to the sensor or something equivalent.
hum, calibrating a transform from the urdf is different because those are joints.
Ok. I searched both keywords in the entire project and I did find that
static
was only really in a few comments not code itself.
great.
I will test it with the softbot now. I'm certain it will give awful results despite not crashing.
Looking forward to hear from it ...
Hi @brunofavs ,
this is solved, right?
Yes
Hey,
Due to now saving transform type in the dataset, in #882 , ALL old datasets are no longer compatible :
A script to convert old datasets into new datasets should be created.