-
Notifications
You must be signed in to change notification settings - Fork 27
Description
Currently, the only way to compress datasets with Silo is for the data producer to use one of Silo's pre-installed filters. There is no way for a data producer to use an arbitrary custom compression filter.
There is a way for a caller to specify arbitrary HDF5 file-access properties but this is only during file create/open (e.g. in DBCreate/DBOpen). To affect compression on individual object basis, we'd have to add it somehow as a DBoption list option. That is doable but requires touching all the places where individual object's optlist options are processed.
The idea would be that a caller could use HDF5 directly creating a dataset creation property list item with their own custom filter in it and then pass the resulting hid_t dcpl id to Silo via this new optlist option. Then, this hid_t would get fed through to the dataset creation operations that occur subsequent to the Silo object DBPutXXX() calls where that optlist option is in effect.