> I have three thoughts:
> - Should we add this to 2.0 at all? It is adding a feature and a 2.1 release
> probably isn't far off.
We have customer using 2.0, who want this feature.
> - I think that certainly in 2.1, this should be default behaviour, as it's
> almost certainly what the user wanted.
Yes, probably you're right.
> - parallel decompression?
I've experimented with parallel decompression, implementing it similar to parallel compression and it didn't gave me any performance gain. It was even slower than thread-per-file implementation. Do you have any specific proposals how to implement parallel decompression?
> I have three thoughts:
> - Should we add this to 2.0 at all? It is adding a feature and a 2.1 release
> probably isn't far off.
We have customer using 2.0, who want this feature.
> - I think that certainly in 2.1, this should be default behaviour, as it's
> almost certainly what the user wanted.
Yes, probably you're right.
> - parallel decompression?
I've experimented with parallel decompression, implementing it similar to parallel compression and it didn't gave me any performance gain. It was even slower than thread-per-file implementation. Do you have any specific proposals how to implement parallel decompression?