Improve file purpose statements
This commit is contained in:
parent
c97b39f9a9
commit
1b887a7e8e
|
@ -1,6 +1,22 @@
|
|||
module Syndicate.Dataspace.Trie.ESOP2016 where
|
||||
-- ESOP 2016 implementation of dataspace tries.
|
||||
-- Includes bug fixes wrt the paper.
|
||||
-- Implementation of dataspace tries, following our ESOP 2016 paper,
|
||||
-- "Coordinated Concurrent Programming in Syndicate" (Tony
|
||||
-- Garnock-Jones and Matthias Felleisen).
|
||||
--
|
||||
-- Includes bug fixes wrt the paper:
|
||||
--
|
||||
-- - combine now has parameters leftEmpty and rightEmpty. In the
|
||||
-- paper, these were missing, and in some cases combine could fail
|
||||
-- to terminate, since it had missing "br(∅)" checks.
|
||||
--
|
||||
-- - we use the smart constructor `tl` throughout, to avoid
|
||||
-- constructing `Tl` atop an empty trie. In the paper, this can
|
||||
-- happen in the definition of `get` when `get(h,★)` is the empty
|
||||
-- trie, but σ=<< and no mapping for σ exists in h.
|
||||
--
|
||||
-- Also, there are problems with the algorithm as described; it is
|
||||
-- roughly correct, but does not collapse away as much redundancy as
|
||||
-- it could. These problems are remedied in ESOP2016v2.hs.
|
||||
|
||||
import Prelude hiding (null, seq)
|
||||
import qualified Data.Map.Strict as Map
|
||||
|
|
|
@ -1,6 +1,6 @@
|
|||
module Syndicate.Dataspace.Trie.ESOP2016v2 where
|
||||
-- Close to the ESOP 2016 implementation of dataspace tries, but takes
|
||||
-- a step toward efficiency by using collapse/update instead of dedup/distinct.
|
||||
-- Builds on ESOP2016.hs, but collapses additional redundancy from
|
||||
-- combine results by using collapse/update instead of dedup/distinct.
|
||||
|
||||
import Prelude hiding (null, seq)
|
||||
import qualified Data.Map.Strict as Map
|
||||
|
|
Loading…
Reference in New Issue