#16093: mkPluginUsage: file not found -------------------------------------+------------------------------------- Reporter: lerkok | Owner: (none) Type: bug | Status: new Priority: normal | Milestone: Component: Compiler | Version: 8.6.3 Resolution: | Keywords: Operating System: Unknown/Multiple | Architecture: | Unknown/Multiple Type of failure: None/Unknown | Test Case: Blocked By: | Blocking: Related Tickets: | Differential Rev(s): Wiki Page: | -------------------------------------+------------------------------------- Comment (by darchon): As a workaround you can run ghci with `-fobject-code`, and for haddock add the `--optghc="-fobject-code"` (you won't need the TemplateHaskell langauge pragma anymore either). Anyhow, this is a regression I introduced when improving/fixing the recompilation logic: https://github.com/ghc/ghc/commit/52065e95c6df89d0048c6e3f35d6cc26ce8246f9 So to determine whether we need to recompile a module we need to track its "inputs", which includes the plugins the the module was compiled with. We decided to use the binary objects, `.so/.a` for packaged plugins, `.o` for local plugins, because they can only give false positives to determine whether the functionality of the plugin changes (`.hi` files can give false negatives). I guess I need to rethink how to handle local plugins... -- Ticket URL: http://ghc.haskell.org/trac/ghc/ticket/16093#comment:8 GHC http://www.haskell.org/ghc/ The Glasgow Haskell Compiler