Ship WB problems and it's future in core (0.18)

[chrisb]

Can you tell from using very simple data which one is right? I would guess that python3 has received improvements. If the results are better there would be no need to search further.

[thschrader]

@looo:
Is it possible for you to add ship-weights (ship, engines, tanks, see tutorial)
and plot the stability curve?
Please see here:
https://forum.freecadweb.org/viewtopic. ... 2&start=10
Thomas

[looo]

Can you tell from using very simple data which one is right? I would guess that python3 has received improvements. If the results are better there would be no need to search further.

I don't know which one is the right one. I compared the data from the table and it looks like there are only differences in the last row. The plots are scaled so it looks like all data is different, but it's only the last one which differs.

Maybe some integer division is involved...

Is it possible for you to add ship-weights (ship, engines, tanks, see tutorial)
and plot the stability curve?

I created a tank, but wasn't able to add point-weights.

[triplus]

Btw.: @triplus What about adding external plot and ship to the next py3-appimage? I think it's time for some testing.

Shouldn't that already work by using the PIP-Tools?

[looo]

Shouldn't that already work by using the PIP-Tools?

yes you are right. But I guess forcing it to be included will give more testers. But I guess such changes are not good for the update-mechanism... So yes tell the testers to use the pip-tools.

[looo]

regarding the difference of the plots:
The problem is caused by rounding some values. The bounding box which is used for the computation sets it's values with a little function which uses the default values for the number of decimals. Because of a problem I have encountered previously there were some changes made by me to not use the rounding for this part of the computation. So I guess the plot on the right (py3) is more correct and maybe there are more issues like these in the code.

[triplus]

But I guess forcing it to be included will give more testers.

I somehow doubt that.

But I guess such changes are not good for the update-mechanism... So yes tell the testers to use the pip-tools.

There i guess still is a problem of not enough users using Py3 builds. And therefore solutions such as PIP-Tools are still limited to rather small potential group of end users.

regarding the difference of the plots:
The problem is caused by rounding some values. The bounding box which is used for the computation sets it's values with a little function which uses the default values for the number of decimals. Because of a problem I have encountered previously there were some changes made by me to not use the rounding for this part of the computation. So I guess the plot on the right (py3) is more correct and maybe there are more issues like these in the code.

That explains it.

[Kunda1]

Can I assign issue #3607 to you?
(if so do you have an account on the tracker? and if so but it's not a 'Developer' account, can you PM @kkremitzki to elevate your privileges?)

[Kunda1]

I've opened a ticket on https://github.com/FreeCAD/freecad.ship/issues/6 to track issue #3607