Uniques are only fixed per-run of the compiler. So when DsMeta (which uses those Names) runs, it generates Core with Names in it (e.g. Language.Hasekell.TH.Syntax.returnQ. When we go to Language/Hasekell/TH/Syntax.hi interface file, and read it in, we'd better use the same unique for return that DsMeta generated. But DsMeta only runs in stage2.
So I think it's fine. Simon | -----Original Message----- | From: ghc-devs [mailto:[email protected]] On Behalf Of | Richard Eisenberg | Sent: 15 November 2013 14:57 | To: [email protected] Devs | Subject: templateHaskellNames | | Hi devs, | | I happened to be looking through the code dealing with known names, and | I'm a little confused about templateHaskellNames. In particular, I'm | confused why it's included in knownKeyNames only when GHCI is defined. | (See lines 196-198 of HscMain.) It's the stage 1 compiler that compiles | the TH library, when the known key names are so important. Couldn't this | cause very strange behavior if some of the numbers are changed? Or, am I | misunderstanding something here? | | This is certainly a minor problem, but it could give someone down the | road a headache if it were to happen. | | Thanks, | Richard | _______________________________________________ | ghc-devs mailing list | [email protected] | http://www.haskell.org/mailman/listinfo/ghc-devs _______________________________________________ ghc-devs mailing list [email protected] http://www.haskell.org/mailman/listinfo/ghc-devs
