[PATCH 0/3] extindex: dedupe support, + gc fix

[Eric Wong <e@80x24.org>]

I'm still not sure how the duplicates got into my extindices;
but the problem doesn't seem reproducible at the moment so
maybe the original bug was fixed.

Since there's already dedupe failures from past indexing, the
--dedupe switch here should help us get rid of them.  It's only
lightly tested, but it seems to be working.

There's also a minor fix for --gc, too.

Eric Wong (3):
  eml: relax warn_ignore regexps for current Email::Address::XS
  extindex: implement --dedupe to fix old extindices
  extindex: --gc: avoid SQLite lock conflict on shard cleanup

 lib/PublicInbox/Eml.pm            |  4 +-
 lib/PublicInbox/ExtSearchIdx.pm   | 96 +++++++++++++++++++++++++++++++
 lib/PublicInbox/OverIdx.pm        | 20 +++++++
 lib/PublicInbox/SearchIdxShard.pm |  5 +-
 script/public-inbox-extindex      | 13 ++++-
 t/extsearch.t                     | 11 ++++
 6 files changed, 142 insertions(+), 7 deletions(-)