We already catch errors when opening the files, so we'll catch if we try to open a file that's not valid HDF5.
This speeds up opening a run by about 5% in my tests. That's a small gain, but it's low-hanging fruit - neither us nor the users need to do anything clever to use it. I'm thinking about how to get bigger speed improvements.
We already catch errors when opening the files, so we'll catch if we try to open a file that's not valid HDF5.
This speeds up opening a run by about 5% in my tests. That's a small gain, but it's low-hanging fruit - neither us nor the users need to do anything clever to use it. I'm thinking about how to get bigger speed improvements.