OK. I think I may have answered my own question with: None of the Above
(or Below, actually, with apologies to Dscho).
It appears that the slowness comes from a very specific clash between
the ZipOutputStream and the file sharing protocol on the server that I'm
trying to write to. If I write the file locally, it happens in no time
flat, give or take a few msec. Easy enough to work around, but an
interesting lesson in where rate-limiting steps can pop up unexpectedly
in an automated workflow.
Bill
On 4/18/11 10:29 AM, Bill Mohler wrote:
> Has anyone else noticed how extremely long it takes to save a zip file
> of ROIs from the ROI manager? I'm seeing times like 6 minutes for 477kb
> of data, 13 minutes for 850 kb... Is this just the slowness of the zip
> algorithm, or is there something intrinisically slow within the
> structure of an ROI being written to disk.
> Or is there something accidental and recursive going on (like rezipping
> the file after each of the several thousand ROIs in a busy ROI manager
> are written? Or is it making calls to the ImagePlus associated with the
> ROIs while writing each one (might make sense, since I'm using a
> VirtualStack to create the ROIs)?
>
> Just wondering if anyone else has noticed and has an explanation. Seems
> like it could be sped up dramatically.
>
> Bill
>
Locked
