If we encounter an IO error while reading packets, we don't want the current behavior of just ending the indexing or decoding here and treating the file as truncated - we want to hard fail so it is apparent there has been an IO issue.
This is helpful in e.g. networked situations.
Note
We ($dayjob) have been using a patch similar to this internally for quite a while, so it should be pretty resilient. I finally got around to cleaning it up for submission here.
If we encounter an IO error while reading packets, we don't want the current behavior of just ending the indexing or decoding here and treating the file as truncated - we want to hard fail so it is apparent there has been an IO issue.
This is helpful in e.g. networked situations.
Note
We ($dayjob) have been using a patch similar to this internally for quite a while, so it should be pretty resilient. I finally got around to cleaning it up for submission here.