syncadncrunch logic at the end of the acquisition on slow connection

[mkrumin]

Again, related to systems with slow connection between ScanImage and processing machines.
According to syncandcrunch logic, stitching starts as soon as there is FINISHED file in the root folder of the sample.
When the data syncing lags behind acquisition it means that stitching starts when not all the raw data for the first channel (blue) is on the processing machine yet, which results in different sizes of the downsampled stacks (stitchit.sampleSplitter fails on loading them) and different number stitched sections for different channels ( blue < green < red)

[raacampbell]

I guess when the FINISHED file appears we could ask it to stop the background rsync and do a final rsync that blocks until everything is copied. Then the stitching starts after that. How would that be?