When you save an "extract block" filter in a catalyst script, the selected blocks for extraction are saved by an index (which appears to be generated as/during a depth-first traverse of the multiblock datastructure).

Unless the data to which the catalyst script is applied has an identical block structure, the extract block filter will fail. User expectations are such that if they specify an extract block filter by clicking the names of the blocks in the gui, the extract block filter will extract the blocks with those names regardless of the block structure which the catalyst script is applied to.

In my own python scripts I have gotten around this by doing the search on the block structure myself in python and finding the required indexes that way before creating the extract block filter. However, I think the appropriate interface for the filter would be to specify included/exclued blocks by block name, not by index. I realize that in some cases blocks might not have names, but we could either allow both methods of specifying block filters, or we could set a default name of a string equal to the block index in cases where no name has been supplied.