#15192: Refactor of Coercion -------------------------------------+------------------------------------- Reporter: ningning | Owner: (none) Type: task | Status: new Priority: normal | Milestone: Component: Compiler | Version: Resolution: | Keywords: Operating System: Unknown/Multiple | Architecture: | Unknown/Multiple Type of failure: None/Unknown | Test Case: Blocked By: | Blocking: Related Tickets: | Differential Rev(s): Phab:D4747 Wiki Page: | -------------------------------------+------------------------------------- Comment (by ningning): - Overall changes so far: 1. Removed {{{Refl r ty}}} and {{{CoherenceCo}}} 2. Introduced {{{Refl ty}}} and {{{GRefl r ty MCoercion}}} - {{{Refl ty :: ty ~n ty}}}, note that {{{Refl ty}}} is always nominal. - {{{GRefl r ty MRefl :: ty ~r ty}}}. If {{{r == Nominal}}}, {{{use Refl}}}. - {{{GRefl r ty (MCo co) :: ty ~r ty |>co}}}. 3. Replaced original {{{Refl Nominal ty}}} with {{{Refl ty}}}. 4. Given {{{g1 :: s ~r t}}}, to construct {{{s |> g2 ~r t}}} we used {{{CoherenceCo g1 g2}}}. It is now replaced with {{{Sym (GRefl r s g2) ; g1}}}. Similar for {{{s ~ t |> g3}}}. 5. It turns out that the explicit patten match in {{{homogenise_result}}} in TcFlatten triggers some optimization of GHC and improves the performance. However it is not useful in master branch. 6. Added a small regression for T9872d (the added number is the allocation of current master on T9872d). 7. Added note about {{{flatten_exact_fam_app_fully}}} performance in TcFlatten. - Performance summary 1. This patch intends to improve the overall performance about coercions. 2. It does perform better in all cases under perf/compiler, except T9872b (0.6%), T9872d(3.7%), and T14683(0.02%). 3. It failed T9872d, thus we added a small regression. 4. It seems to perform better to compile large packages, e.g. Cabal. - TODO: Further analysis of the performance: a step-by-step replay of the refactor. -- Ticket URL: http://ghc.haskell.org/trac/ghc/ticket/15192#comment:13 GHC http://www.haskell.org/ghc/ The Glasgow Haskell Compiler