I'd bet that this is already on your list, Tim, so this is just a reminder so that we don't forget when prepping for release: reports generation should in principle be generated from counts and/or loadfiles, by a distinct process invoked after loadfile generation ... rather than by function call within the same process. I think this is what sample_report.py embodies, but there's some (perhaps leftover) code in
create_loadfile which imports report.py, which in turn also imports matplotlib ... and immediately breaks on my virginal homebrew-ed Python setup.
I'd bet that this is already on your list, Tim, so this is just a reminder so that we don't forget when prepping for release: reports generation should in principle be generated from counts and/or loadfiles, by a distinct process invoked after loadfile generation ... rather than by function call within the same process. I think this is what sample_report.py embodies, but there's some (perhaps leftover) code in create_loadfile which imports report.py, which in turn also imports matplotlib ... and immediately breaks on my virginal homebrew-ed Python setup.