RE: cvs commit: hugs98 Readme hugs98/docs hugs.1.in hugs98/docs/users_guide hugs_ghc.sgml introduction.sgml running.sgml hugs98/src Makefile.in input.c opts.c opts.h parser.y hugs98/src/unix hugs-package.in test_libraries hugs98/tests ...
ross 2003/09/18 11:02:25 PDT
Modified files: . Readme docs hugs.1.in docs/users_guide hugs_ghc.sgml introduction.sgml running.sgml src Makefile.in input.c opts.c opts.h parser.y src/unix hugs-package.in test_libraries tests testScript Log: Remove the obsolete dlet/with syntax, and the +/-W option.
Should we do this in GHC too? For 6.2 maybe?
I'd love to see it stay, because it is syntactically oh-so-much-nicer than `let' in practice.
However, having lost that battle previously (albeit for the wrong reasons, of course ;-), I can't think of a good reason to keep it. And besides, it bugs Sven, so lets take it out of GHC (for 6.2) as well.
Ideally we'd like to keep everyone happy. Sometimes that's not possible, and this is one of those cases. Now, the 'with' keyword isn't doing anyone any harm where it is apart from adding a bit of complexity to the parser, and I'm not really bothered by that, so I really don't mind leaving it in if you're still using it. I think the long-term outlook for 'with' is bleak, though. Cheers, Simon
Ideally we'd like to keep everyone happy. Sometimes that's not possible, and this is one of those cases. Now, the 'with' keyword isn't doing anyone any harm where it is apart from adding a bit of complexity to the parser, and I'm not really bothered by that, so I really don't mind leaving it in if you're still using it.
Except that it conflicts with an identifier used in a standard library. Could we fix this using the same ugly hack we use for the 'as' keyword? That is, treat it as a keyword only when used in the right context. I'm not sure of the answer - seems like it would require extra lookahead to decide which way 'with' is being used. -- A
participants (2)
-
Alastair Reid -
Simon Marlow