David Roundy wrote: > On Fri, Sep 26, 2008 at 07:22:39PM +0100, Simon Marlow wrote: >> David Roundy wrote: >>> On Thu, Sep 25, 2008 at 4:02 PM, David Roundy <[EMAIL PROTECTED]> wrote: >>>> On Thu, Sep 25, 2008 at 11:06:23AM -0700, Dmitry Kurochkin wrote: >>>>> Is there any reason not to use System.Directory.canonicalizePath in >>>>> ioAbsolute? It seems to do just what we need. >>> .... >>>> I'm also still not entirely comfortable with the decision to rely on >>>> the filepath package and break compatibility with ghc 6.4, so I'd >>>> rather keep the number of uses down, for the moment. >>> Ooops. I failed to pay attention to where canonicalizePath was >>> located, so this is irrelevant. But the issue still holds that its >>> behavior is undocumented, and so the source code would have to be >>> examined before we could could seriously consider this move. Note >>> that two functions above canonicalizePath in the haddock page is a >>> function whose description doesn't match its behavior. >> You mean the atomicity of renameFile? We fixed that in GHC 6.10.1, FWIW. > > Really? How did you do it? Is it truly atomic? I thought that wasn't > possible under windows.
Well, ok, having just read the ticket again, it seems there's no actual atomicity guarantee provided by the underlying Win32 API. Here's the ticket we fixed in 6.10.1: http://hackage.haskell.org/trac/ghc/ticket/2298 Cheers, Simon _______________________________________________ darcs-users mailing list [email protected] http://lists.osuosl.org/mailman/listinfo/darcs-users
