[Haskell-cafe] Why purely in haskell?

[Henning Thielemann]

On Thu, 10 Jan 2008, David Roundy wrote:

> On Thu, Jan 10, 2008 at 08:10:57PM +0000, Sebastian Sylvan wrote:
> > On Jan 10, 2008 8:06 PM, Ketil Malde <ketil+haskell at ii.uib.no> wrote:
> > > "David Roundy" <daveroundy at gmail.com> writes:
> > >
> > > >> > I just want to point out that unsafePerformIO is at the core of the
> > > >> > (safe) bytestring library.  As SPJ et al pointed out, this is crucial
> > > >> > functionality, and is only unsafe if unsafely used.
> > >
> > > >> In Modula-3 modules using hacks must be explicitly marked as UNSAFE. See
> > > >>   http://www.cs.purdue.edu/homes/hosking/m3/reference/unsafe.html
> > > >>  Maybe this is also an option for Haskell?
> > >
> > > > I don't think this is a good idea.
> > >
> > > I think the point is (should be) to mark functions unsafe when they
> > > may be unsafe to /use/,
> >
> > I think using the IO monad for this works well...
>
> Would you suggest moving head and tail into the IO monad?

I'm afraid we are talking about different notions of 'safe'. Modula-3's
'safe' means no "segmentation fault", but program abortion due to ASSERT
is still allowed. Ported to Haskell this means: 'head' and 'tail' are
safe, but not total. I've seen function definitions like 'safeHead', that
I would have named 'maybeHead'. For running untrusted code this means: I
think it is ok that the program aborts with an error or runs into an
infinite loop and must be terminated after a time-out, but it is not ok,
that it overwrites some memory area or deletes some files, because
unsafePerformIO was invoked.