On Mon, 12 Dec 2005, Duncan Coutts wrote: > On Mon, 2005-12-12 at 09:52 -0800, Donn Cave wrote: > > > Of course there's a risk that the authors of ghc may notice that > > we're doing this and come up with a way to thwart it, but it seems > > to me that between interfering with legitimate applications and not > > working reliably anyway, there'd be a case for letting go of this > > notion altogether. > > Except that it is specified to work this way in the Haskell Report. > > Instead of changing the semantics of the standard libraries what you > want is a back door to allow the alternative semantics that you want. > Even better would be to provide a supported api for opening files > without locking.
Actually I think I would prefer that the semantics change. A work-around would be good, and its existence would help publicize the problem with openFile, but changing openFile would be better for everyone, I suspect. (Apparently it may have been so specified in Haskell 98, up to April 2001?) Donn Cave, [EMAIL PROTECTED] _______________________________________________ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe