Details
-
Type:
Improvement
-
Status: Won't Fix
-
Resolution: Done
-
Fix Version/s: None
-
Component/s: afw
-
Labels:None
-
Team:Data Release Production
Description
The LSST camera will apparently write 18 bits of data per pixel, so we may end up storing images as uint32. We might as well be prepared by instantiating them.
This affects afw::math::Statistics and our SWIG interfaces as well. Note that we presently use these one-character suffixes in our SWIG wrappers: U=uint16, L=uint64. I'm not sure what we'd use for uint32, but if you want it to be obvious to the users I strongly suggest changing to U16, U32 and U64 for those cases where there is more than one option (hence not masks, which should stay U as long as we only have one mask pixel type).
Attachments
Activity
Calling this Won't Fix, as I think we've settled into using signed 32-bit ints in places where unsigned 32-bit ints might be more natural but 31 unsigned bits are sufficient.
I assume we no longer have any plans for this given that we are taking data with LSSTCam and immediately converting to floating point when we read the raw?