I’ve been working on a visualization tool for a while, in Jupyter notebooks on Windows and macOS. It wasn’t completely without issues, but for the most part it worked fine. I use Plotly to generate maps and export map images to PNG. Orca is somehow involved in that process as a dependency, which with Jupyter on win/mac works mostly fine, it’s all handled automatically.
But my ultimate goal was to put that app on a Linux machine and run it there on a regular schedule. On Linux I had nothing but trouble with Orca - see this bug report:
https://github.com/plotly/orca/issues/330
I could only make it work for really simple cases, like a test script that only makes one map. If I try to do anything more complex than that, it fails. I don’t even do multiprocessing, it’s a single process feeding requests sequentially to Orca. And when it fails, there’s a random aspect to that - sometimes it quits without making a single image, other times I get “unable to communicate with the orca server…” after a handful of requests.
It’s very frustrating and I’ve wasted a lot of time debugging this thing.
So here’s my question: perhaps running Orca on Linux is not really something that would be recommended? Am I going completely against the grain here? If so, I’ll have to re-think my whole approach.
Thanks!