[Bug #116679] -fglasgow-exts for -syslib lang

[Simon Peyton-Jones]

I've had this lying in my inbox for some while.

The Right Thing is to make the Haskell98 checks only for
*source* code, not for interface-file code.  I've done this in
the HEAD, so the problem you rightly identify will go away in due
course.

Simon

| -----Original Message-----
| From: noreply@sourceforge.net [mailto:noreply@sourceforge.net]
| Sent: 12 October 2000 03:46
| To: chak@cse.unsw.edu.au; noreply@sourceforge.net; cvs-ghc@haskell.org
| Subject: [Bug #116679] -fglasgow-exts for -syslib lang
|=20
|=20
| Bug #116679, was updated on 2000-Oct-11 19:46
| Here is a current snapshot of the bug.
|=20
| Project: The Glasgow Haskell Compiler
| Category: hslibs/lang
| Status: Open
| Resolution: None
| Bug Group: None
| Priority: 5
| Summary: -fglasgow-exts for -syslib lang
|=20
| Details: Toying around with the module IArray from -syslib=20
| lang, Gabi and I found that we need to give ghc=20
| -fglasgow-exts, because IArray is a multi-parameter type class.
|=20
| This is a problem, because a user less familar with ghc, will=20
| probably not be able to infer from some strange error message=20
| whose position is in ArrayBase.hi what the problem is.  This=20
| suggests that the driver should automatically add=20
| -fglasgow-exts when -syslib lang is specified.
|=20
| The problem with this strategy is, however, that also when=20
| somebody imports components of -syslib lang that do not need=20
| extensions, the flag will be used.  So, maybe we should split=20
| lang into lang-h98 and lang-exts or so.
|=20
| At the very least, the documentation for IArray should=20
| mention this caveat.
|=20
|=20
| For detailed info, follow this link:
| http://sourceforge.net/bugs/?func=3Ddetailbug&bug_id=3D116679&grou
p_id=3D8032

_______________________________________________
Cvs-ghc mailing list
Cvs-ghc@haskell.org
http://www.haskell.org/mailman/listinfo/cvs-ghc