Re: [PATCH 06/11] Better error handling in compress_all()

Marco Costalba [off-list ref] writes:

Also let the caller to xmalloc() the buffer
int compress_start()

This is meant to be an improvement for [01/11] and I think
should be done from the beginning by squashing into it.

Haven't looked at the decompression side yet.  Help in reviewing
this series from others are appreciated.