#1480: Template Haskell should allow reification of modules ----------------------------+---------------------------------------------- Reporter: igloo | Owner: Type: feature | Status: closed request | Milestone: 7.8.1 Priority: highest | Version: 7.7 Component: | Keywords: Template Haskell | Architecture: Unknown/Multiple Resolution: fixed | Difficulty: Moderate (less than a day) Operating System: | Blocked By: Unknown/Multiple | Related Tickets: Type of failure: | None/Unknown | Test Case: | Blocking: 8398 | ----------------------------+---------------------------------------------- Comment (by errge): I have to experiment a bit more around this, but at a first glance, I see this in ```HscTypes.lhs```: {{{ -- | Records modules that we depend on by making a direct import from data Usage -- | Module from another package = UsagePackageModule { usg_mod :: Module, -- ^ External package module depended on usg_mod_hash :: Fingerprint, -- ^ Cached module fingerprint usg_safe :: IsSafeImport -- ^ Was this module imported as a safe import } -- | Module from the current package | UsageHomeModule { usg_mod_name :: ModuleName, -- ^ Name of the module usg_mod_hash :: Fingerprint, -- ^ Cached module fingerprint usg_entities :: [(OccName,Fingerprint)], -- ^ Entities we depend on, sorted by occurrence name and fingerprinted. -- NB: usages are for parent names only, e.g. type constructors -- but not the associated data constructors. usg_exports :: Maybe Fingerprint, -- ^ Fingerprint for the export list we used to depend on this module, -- if we depend on the export list usg_safe :: IsSafeImport -- ^ Was this module imported as a safe import } -- ^ Module from the current package -- | A file upon which the module depends, e.g. a CPP #include, or using TH's -- 'addDependentFile' | UsageFile { usg_file_path :: FilePath, -- ^ External file dependency. From a CPP #include or TH -- addDependentFile. Should be absolute. usg_file_hash :: Fingerprint -- ^ 'Fingerprint' of the file contents. -- Note: We don't consider things like modification timestamps -- here, because there's no reason to recompile if the actual -- contents don't change. This previously lead to odd -- recompilation behaviors; see #8114 } deriving( Eq ) -- The export list field is (Just v) if we depend on the export list: -- i.e. we imported the module directly, whether or not we -- enumerated the things we imported, or just imported -- everything -- We need to recompile if M's exports change, because -- if the import was import M, we might now have a name clash -- in the importing module. -- if the import was import M(x) M might no longer export x -- The only way we don't depend on the export list is if we have -- import M() -- And of course, for modules that aren't imported directly we don't -- depend on their export lists }}} As you can see, UsagePackageModule doesn't have a usg_exports field, only UsageHomeModule. So I can't decide about directness of the import based on usg_exports alone. Maybe we should make this info more explicit in Usages if we want to depend on it for TH reification. -- Ticket URL: http://ghc.haskell.org/trac/ghc/ticket/1480#comment:21 GHC http://www.haskell.org/ghc/ The Glasgow Haskell Compiler