On Tue, Oct 6, 2015 at 2:01 PM, Roelof Wobben wrote:

but why this part :

x (x + y) (x - (y + 1))

That is buggy. I was trying to quickly create three arguments that are guaranteed to be different by some random amount, but what I wrote doesn't always work (eg: if y = 0, then the first two arguments are the same, if y is -1, the first and 3rd are the same, if x is MAX_INT, then it may or may not be the same as other arguments, depending on Y, etc...) The predicate guard should work, though it may take longer to run. --Rogan

Roelof

Op 6-10-2015 om 22:31 schreef Rogan Creswick:

...

On Tue, Oct 6, 2015 at 1:24 PM, Roelof Wobben wrote:

...

Hello,

I have written a function to test if three numbers are the same.

Now I want to test my functions by using quickCheck.

So I thought to test first if all three are the same.

but how can I tell quickcheck that all numbers are the same.

Just have quickcheck generate one number, and use it for all three arguments, eg:

prop_allSame :: Int -> Bool prop_allSame x = yourFn x x x

then test the other cases:

prop_different :: Int -> Int -> Int -> Property prop_different x y z = x /= y && y /= z && x /= z ==> not $ yourFn x y z

That's probably OK for ints, but generally the guard in that style isn't a great solution, since quickcheck will just keep generating inputs until the predicate is satisfied. That can take a while, so you could also manually offset the first input in various ways:

prop_different :: Int -> Int -> Bool prop_different x y = not $ yourFn x (x + y) (x - (y + 1)) -- I put a +1 in here to avoid returning true where y = 0,

There are probably other ways to mutate a randomly generated input for your problem that may work better, but that's the general idea.

--Rogan

...

and second question : hwo do the test likeif two numbers are the same.

I ask this because I try to solve a exercise of the programming Haskell book, IM have read chapter 3 now.

Roelof

_______________________________________________ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://mail.haskell.org/cgi-bin/mailman/listinfo/haskell-cafe

----- Geen virus gevonden in dit bericht. Gecontroleerd door AVG - www.avg.com Versie: 2015.0.6140 / Virusdatabase: 4435/10768 - datum van uitgifte: 10/06/15

_______________________________________________ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://mail.haskell.org/cgi-bin/mailman/listinfo/haskell-cafe

On Wed, Oct 7, 2015 at 12:13 AM, Roelof Wobben wrote:

I know that if three numbers are the same then the outcome will be false. The same if two numbers are the same.

Can I not do something like this :

prop_different x y z = not( (x /= y && y /= z && x /= z))

to test all cases at once.

That looks like it will work (I'm assuming you're actually calling your function in the property and comparing with that boolean expression); is that the same way your function is implemented? --Rogan

Roelof

Op 7-10-2015 om 00:28 schreef Rogan Creswick:

...

On Tue, Oct 6, 2015 at 2:01 PM, Roelof Wobben wrote:

...

but why this part :

x (x + y) (x - (y + 1))

That is buggy. I was trying to quickly create three arguments that are guaranteed to be different by some random amount, but what I wrote doesn't always work (eg: if y = 0, then the first two arguments are the same, if y is -1, the first and 3rd are the same, if x is MAX_INT, then it may or may not be the same as other arguments, depending on Y, etc...)

The predicate guard should work, though it may take longer to run.

--Rogan

...

Roelof

Op 6-10-2015 om 22:31 schreef Rogan Creswick:

...

On Tue, Oct 6, 2015 at 1:24 PM, Roelof Wobben wrote:

...

Hello,

I have written a function to test if three numbers are the same.

Now I want to test my functions by using quickCheck.

So I thought to test first if all three are the same.

but how can I tell quickcheck that all numbers are the same.

Just have quickcheck generate one number, and use it for all three arguments, eg:

prop_allSame :: Int -> Bool prop_allSame x = yourFn x x x

then test the other cases:

prop_different :: Int -> Int -> Int -> Property prop_different x y z = x /= y && y /= z && x /= z ==> not $ yourFn x y z

That's probably OK for ints, but generally the guard in that style isn't a great solution, since quickcheck will just keep generating inputs until the predicate is satisfied. That can take a while, so you could also manually offset the first input in various ways:

prop_different :: Int -> Int -> Bool prop_different x y = not $ yourFn x (x + y) (x - (y + 1)) -- I put a +1 in here to avoid returning true where y = 0,

There are probably other ways to mutate a randomly generated input for your problem that may work better, but that's the general idea.

--Rogan

...

and second question : hwo do the test likeif two numbers are the same.

I ask this because I try to solve a exercise of the programming Haskell book, IM have read chapter 3 now.

Roelof

_______________________________________________ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://mail.haskell.org/cgi-bin/mailman/listinfo/haskell-cafe

----- Geen virus gevonden in dit bericht. Gecontroleerd door AVG - www.avg.com Versie: 2015.0.6140 / Virusdatabase: 4435/10768 - datum van uitgifte: 10/06/15

_______________________________________________ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://mail.haskell.org/cgi-bin/mailman/listinfo/haskell-cafe

----- Geen virus gevonden in dit bericht. Gecontroleerd door AVG - www.avg.com Versie: 2015.0.6140 / Virusdatabase: 4435/10768 - datum van uitgifte: 10/06/15

_______________________________________________ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://mail.haskell.org/cgi-bin/mailman/listinfo/haskell-cafe

On Wed, Oct 7, 2015 at 10:01 AM, Roelof Wobben wrote:

Op 7-10-2015 om 18:32 schreef Rogan Creswick:

...

On Wed, Oct 7, 2015 at 12:13 AM, Roelof Wobben wrote:

...

I know that if three numbers are the same then the outcome will be false. The same if two numbers are the same.

Can I not do something like this :

prop_different x y z = not( (x /= y && y /= z && x /= z))

to test all cases at once.

That looks like it will work (I'm assuming you're actually calling your function in the property and comparing with that boolean expression); is that the same way your function is implemented?

--Rogan

Unfortanally yes but I do not know another way to test it. Im only a beginner which has read 2 chapters of the Crafts book.

Then I'd suggest using the quickcheck property guards ( `==>`) to only generate inputs that are known to be equal, or not equal and test separately in a few different properties. --Rogan

Roelof

_______________________________________________ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://mail.haskell.org/cgi-bin/mailman/listinfo/haskell-cafe

On Wed, Oct 7, 2015 at 11:33 AM, Ozgur Akgun wrote:

This is not true, ==> still does generate and test.

I could have phrased that better. Sure, it generates values, but it will never provide them to your test, so your logic is simpler.

Prelude> import Test.QuickCheck Prelude Test.QuickCheck> let f x = x == 13 ==> mod x 2 == 1 Prelude Test.QuickCheck> quickCheck f *** Gave up! Passed only 6 tests.

That's unlikely to be a problem given the test inputs in the original question, but yes, it is certainly an issue to know about. Roelof - The issue Ozgur is demonstrating is that the ==> operator will just discard test inputs that were generated but which did not satisfy the predicate (here that x == 13). QuickCheck will eventually stop trying to generate inputs (you can tune this), and produce something like the above result. --Rogan

Also try verboseCheck to see what values are generated.

Ozgur

Op 8-10-2015 om 01:16 schreef Richard A. O'Keefe:

You can generate two unequal integers by doing

prop_twoDifferent :: Int -> Int -> Bool prop_twoDifferent x y = yourFn2 x y' where y' = if y >= x then y+1 else y

Letting the number of distinct Int values be N, there are N*N pairs of Ints but only N*(N-1) pairs of *different* Ints, so this technique will generate some x y' pairs more often than others. In fact you will get (x,minBound::Int) with probability 2/N and (x,y>minBound) with probability 1/N. The advantage of using a guard is that the probabilities come out right (it's the "rejection method", in statistical parlance).

You can use that method to generate x x y x y x y x x

You can iterate the same idea to generate triples.

----- Geen virus gevonden in dit bericht. Gecontroleerd door AVG - www.avg.com Versie: 2015.0.6140 / Virusdatabase: 4435/10775 - datum van uitgifte: 10/07/15

Thanks all. Enough to study and try. Roelof