Open slactjohnson opened 1 year ago
Repository is here: https://github.com/pcdshub/l2si-beam-transport-system-displays
PyDM v1.11.1 to v1.12.0 change apparently caused this:
PyDMDrawingLine's default orientation was flipped 90 degrees at some point
edit: the default orientation was changed from 45 degrees to 0 degrees, but that should have only affected lines that were using "default/unset" rotation
Even fixing the rotation is insufficient: the line doesn't fill the bounding box of the widget. I'll attach a screenshot in a bit.
Here's the relevant pull request with all the details of the changes that must be affecting this: https://github.com/slaclab/pydm/pull/745
My read is that the original behavior was "draw a line from bottom left to top right (or is it top left to bottom right?) inside the bounding box and then rotate it with a default rotation of 45 degrees" while the new behavior is "draw a line from the left side to the right side of the bounding box and then rotate it with a default rotation of 0 degrees"
So to make a long vertical line, you have to make an extremely large bounding box for the line. Here's a rotation=90 deg line with a variety of widget sizes in the designer:
No scaling has been applied to these images; they're a simple crop. Use the designer background grid for reference.
So it draws from the left bounding box to the right and then rotates it around the middle point without extending it?
This smells like a needed PyDM PR
I think you've got the right logic there @ZLLentz
I can cook something up this afternoon, and we could even slap it into our dev env prior to any pydm tagging
The laser hall will be OK for now using an old PyDM, but I think we should get this fixed sooner rather than later regardless
My fix got merged into PyDM already :tada:
After the patch we'd probably need to make modifications to the screen still- we can do that whenever we deem it a good timing.
In the time since ~Tuesday morning and now, something appears to have broken the LAS BTS vacuum screen. See below.
Placing this issue in this repo at @klauer 's request, since it's not really clear where this problem originates.