Whoosh issues python3 (attn Joseph/Christof)

Hi all:

I applied some fixes to the indexer code to make German and Russian
words index correctly. While I was doing that I also reindexed a 145MB
tracker with about 2000 issues using all indexer methods.

When I was reindexing, it looks like I (mistakenly) chose to use
python 2 to run the roundup-admin reindex command.

However the tracker runs under python3.

For native database text indexing, I used sqlite. Since the db doesn't
need to be dumped/restored for the python2 -> 3 migration it worked.

It looks like xapian also worked.

However the whoosh indexed databases threw unicode "can't decode"
exceptions when used.

Does anybody know why? Whoosh is pure python, but I am not sure if
pickling format or something else that whoosh uses in its indexer
changed between 2 and 3.

It looks like reindexing whoosh using python3 and running using
python3 works. So I will recommend (slow) reindexing as part of the
python 3 migration docs. But it would be nice to understand why this
happens and if there is a way to recover without the reindex.

Any quips, comments, evasions, questions or answers?

--
				-- rouilj
John Rouillard
===========================================================================
My employers don't acknowledge my existence much less my opinions.