[ome-devel] Bug 274 : Incremental Import

[Joshua Allen Moore]

Quoting Jason Swedlow <jason at lifesci.dundee.ac.uk>:

>   How much data will be on board when use needs to make the decision to 
> continue or cancel?  Is it really aproblem to have the "preliminary" 
> data on the local filesystem?
> 
> Each single channel image is 1Mb, or roughly 750 x 750??  How many 
> channels???

Sorry my last reply was hasty because I had to get out of the office. (Ever have 
that feeling? :) )

The decision to continue or cancel can be made with a few time steps (I assume) 
but it's more a matter of trying to keep up with the automation software. What 
the guys at EMBL want to avoid--and this bodes well for OME--is programming 
their algorithms multiple times. That is, when they have the OME infrastructure, 
why not use it right off the bat. 

For example, a good deal of the classifiers we at DKFZ will be creating could be 
put "online" to prevent certain classes of images from even being scanned. If we 
program our classifiers against OME, what's the development time to program it 
against the filesystem.

So, no, it's not a matter of filesystem space in the beginning really, it's more 
a human-machine interaction thing, which certainly isn't do-or-die, but could 
prove awfully nice.

Josh.

The file sizes are estimates from our last head-to-head so don't take them too 
literally. We were trying to come up with the bandwidth we need. The channel 
estimate is 3.