#2120: Arrays allow out-of-bounds indexes
----------------------------------+-----------------------------------------
Reporter: amthrax | Owner:
Type: bug | Status: new
Priority: normal | Milestone: 6.12.1
Component: libraries (other) | Version: 6.8.2
Severity: normal | Resolution:
Keywords: | Difficulty: Unknown
Testcase: | Os: Unknown/Multiple
Architecture: Unknown/Multiple |
----------------------------------+-----------------------------------------
Comment (by simonpj):
Triggered by this thread http://www.haskell.org/pipermail/haskell-
cafe/2009-June/063399.html, I had quick look.
There are two range tests under discussion
* One tests every index supplied by the client of the array, against the
original bounds. We should never leave this test out.
* The other tests the `Int` offset computed by `index`, in case the `Ix`
instance for this type is bogus. We can omit this check iff we trust the
instance.
The only safe thing to do (and Haskell is supposed to be a safe language)
is to do both checks, thus (in `GHC.Arr`):
{{{
safeIndex :: Ix i => (i, i) -> Int -> i -> Int
safeIndex (l,u) n i = let i' = index (l,u) i
in if (0 <= i') && (i' < n)
then i'
else error "Error in array index"
}}}
(Note the use of `index` rather than `unsafeIndex`.) To avoid the double
test in the (wildly common) cases of indexing using the (trusted) built-in
instances for `Int`, `(Int,Int)` etc, we could use a RULE to call version
of `safeIndex` that did only one test.
Furthermore, we should improve the "Error in array index" error message.
If we have the first client-oriented test in place, then this second error
can read something like "The index method for an Ix instance returned
offset N, but the array has size M". I don't see how to say ''which''
type, sadly. `Typeable` is not a superclass of `Ix`.
Simon
--
Ticket URL: <http://hackage.haskell.org/trac/ghc/ticket/2120#comment:11>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler_______________________________________________
Glasgow-haskell-bugs mailing list
Glasgow-haskell-bugs@haskell.org
http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs
