Still doing Miranda-style pattern matching?

[Jan-Willem Maessen]

We were quite surprised when a student mnaged to get away with the
following (skeletal) definition:

sumsquares n | n < 0  = error "sumsquares:n < 0"
           n | n >= 0 = n

since according to the Haskell report this either shouldn't parse, or
should parse as two definitions---one for "sumsquares" and one for "n"
(a twiddly bit of the report I didn't cross-reference).  My only guess
is that this is a holdover from the Miranda-like days of Gofer.  It
does confuse students whose code suddenly stops working when given to
a compiler...

Or am I totally misreading pp 46--48 of the report?  (I note that hbc
and our compiler, both of which are at best Haskell 1.3 compliant
around here, parse this the same way as Hugs does---but I'm sure I've
seen complaints about this usage before.)

                                                -Jan-Willem Maessen