It seems like use of a lot of extensions is so obvious from syntax that the compiler is able to suggest the correct pragma to turn on to enable that syntax. In fact the way I know to use most of these pragmas is because the compiler told me about them. So, my suggestion is that in any case where the compiler currently suggests use of a particular pragma in an error message, it should instead turn that pragma on and produce a warning. Does that make sense? -Alex- Simon Peyton-Jones wrote:
Alex,
| Ok, I'm game to default to haskell98 in the presence of ambiguity, but | in most cases the extension involves new syntax and that should be enough.
Trying to compile the program both ways (or 2^n ways) to check for ambiguity sounds like a pretty heavy hammer to crack this nut.
GHC is a Haskell-98 compiler by default, and I think it's a good default. You are arguing to change the default, but that's always an uphill battle because there will be others with good user-based reasons (like yours) for preferring the current default.
Better would be to make it easy to get what you want. It is certainly useful to have a name for a bundle of flags. Currently -O, -O2, and -fglasgow-exts, and -Wall are the only such names.
But we don't currently have a "switch on all extensions" flag, -Xall say. If you'd find that useful, you could put {-# OPTIONS_GHC -Xall #-} at the top of all your modules, and you'd be done. In situations where you didn't want a particular extension (like Duncan's example) you can switch it off selectively with -XNoBangPatterns.
Simon