eliminate single/multi writer distinction
authorJoey Hess <joeyh@joeyh.name>
Wed, 20 Oct 2021 16:24:40 +0000 (12:24 -0400)
committerJoey Hess <joeyh@joeyh.name>
Wed, 20 Oct 2021 16:26:30 +0000 (12:26 -0400)
commitf5b642318dc33f9cee41cac08a276cae37878633
tree080be0b502e6fe36f5b1070269d61c81e220fdc8
parentc47794991cbc9f04bf4bab2bb51da5c1a33d23e2
eliminate single/multi writer distinction

After commit f4bdecc4ecce8bcbb2d8ccbb524ba1f3f76565ba, there is no
longer any distinction between SingleWriter and MultiWriter's handling
of read after write.

Databases that were SingleWriter still have lock files that are used to
prevent multiple writers.

This does make writing to such databases a bit more expensive,
because the MultiWriter code path that is now used opens a second db
connection in order to write to them.
Database/Benchmark.hs
Database/ContentIdentifier.hs
Database/Export.hs
Database/Fsck.hs
Database/Handle.hs
Database/Keys.hs
Database/Queue.hs