Tim Allison created TIKA-4828:
---------------------------------
Summary: Improve embedded object digesting in 4.x
Key: TIKA-4828
URL: https://issues.apache.org/jira/browse/TIKA-4828
Project: Tika
Issue Type: Task
Reporter: Tim Allison
3.x tee'd the hash off the single parse read and never buffered.
This felt complicated to me on mark/reset and exception paths, and I wanted to
digest first, parse second in 4.x. This winds up spilling 4GB to disk for a
small mixed corpus of files where 3.x didn't write anything to disk. On a slow
disk, this causes a performance hit.
Let's see if we can do a better job.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
