Web: thumbs direct?

[wmoore]

Currently we use all the 'Direct' methods to access thumbnails in Blitz Gateway getThumbnail()

• getThumbnailByLongestSideDirect
• getThumbnailForSectionByLongestSideDirect
• getThumbnailDirect
• getThumbnailForSectionDirect

But this causes extra overhead since existing thumbnails may already exist.

In my simple testing, getThumbnailByLongestSide() seemed to perform fine, reusing existing thumbnails when the rendering settings hadn't changed, but creating a new one when they had. "Creating a new one" was observed by way of log statements such as (also see #10337):

2013-02-07 10:55:28,772 INFO  [              loci.formats.FormatHandler] (l.Server-9) Initializing helper readers
2013-02-07 10:55:28,772 INFO  [              loci.formats.FormatHandler] (l.Server-9) Reading additional metadata
2013-02-07 10:55:28,783 INFO  [              loci.formats.FormatHandler] (l.Server-9) Populating metadata
2013-02-07 10:55:28,784 INFO  [              loci.formats.FormatHandler] (l.Server-9) Reading bitmap header
2013-02-07 10:55:28,784 INFO  [              loci.formats.FormatHandler] (l.Server-9) Populating metadata
2013-02-07 10:55:28,789 WARN  [el.enums.handlers.LaserMediumEnumHandler] (l.Server-9) Unknown LaserMedium value 'Rhodamine Green' will be stored as "Other"
2013-02-07 10:55:28,789 WARN  [el.enums.handlers.LaserMediumEnumHandler] (l.Server-9) Unknown LaserMedium value 'TRITC' will be stored as "Other"
2013-02-07 10:55:28,790 WARN  [              loci.formats.FormatHandler] (l.Server-9) Expected positive value for EmissionWavelength; got 0
2013-02-07 10:55:28,790 WARN  [el.enums.handlers.LaserMediumEnumHandler] (l.Server-9) Unknown LaserMedium value 'None' will be stored as "Other"
2013-02-07 10:55:28,790 INFO  [                ome.io.nio.PixelsService] (l.Server-9) Creating BfPixelBuffer: /OMERO/ManagedRepository/system/root/2013-2-6/Users/will/Documents/biology-data/Test-Import-Images/FV1000.oib/aryeh/1h.oib Series: 0

If we use getThumbnailByLongestSideDirect() instead, then we see these log statements every time we request a thumbnail, even if a suitable one already exists.

Carlos, is there something I'm missing (a reason to use *Direct methods) or should we switch them all over to using non-Direct methods?

[wmoore]

Fixed in ​https://github.com/openmicroscopy/openmicroscopy/pull/723