The calculation of number and size of hdf5 cache slots is wrong when data
is 1D, in case of multiframe file, it shouldn't consider the 0 dimension,
as the number of frames is calculated separately.
In the case of non multiframe file, it will try to access the invalid
dimension -1.
The approach taken is checking that we are not in one of those cases
before considering the dimension (this->rank-2), which would be 0 in
a multiframe file, and -1 in a non multiframe file
The symptoms of the problem that this intend to fix depend on many factors
and could be any of the following:
Coverage increased (+0.009%) to 38.908% when pulling c0bd10be9e9cdbaf8348e560d4bf7121b4e5e18c on dls-controls:fix-cache-size into bcaced94ad48b9bb5fdbd1a7f6d9d8be5c416003 on areaDetector:master.
The calculation of number and size of hdf5 cache slots is wrong when data is 1D, in case of multiframe file, it shouldn't consider the 0 dimension, as the number of frames is calculated separately. In the case of non multiframe file, it will try to access the invalid dimension -1. The approach taken is checking that we are not in one of those cases before considering the dimension (this->rank-2), which would be 0 in a multiframe file, and -1 in a non multiframe file
The symptoms of the problem that this intend to fix depend on many factors and could be any of the following: