Ralph Johnson wrote:
>> Did you delete the tests AND the code that it tests? Just removing the
>> tests doesn't really help much if the code that it tests is still there.
>> I'm sure you agree with this, so that is why I'm a bit stumped.
> 
> You completely miss the point.  Leaving broken tests in the image does
> not cause the tests to be fixed.  Broken tests have been in the image
> for years and they did not help.
> 
> There are a lot of bugs in Squeak.  Squeak is, in general, of
> "experimental" quality.  It is amazing to me that people actually
> develop commercial products with it, but they do.  I would like to
> raise the quality of Squeak.  Part of this is making sure that there
> is a regression test suite that actually works, and preventing people
> from breaking it.  There are lots of other parts, of course.  The
> image should be smaller, it should have more tests, more bugs should
> be fixed, there should not be any unimplemented messages.  No one
> thing will improve quality, but a lot of little things will.
> 
> In general, it is not easy to find the code that a test exercises.
> What you are asking does not  make sense.
> 
> Please run the tests in 3.9, look at the ones that are broken, and try
> to figure out what code is bad and should be deleted.  It will be an
> enlightening experience.

I agree with you that there are a lot of bugs in squeak and they should
be fixed. I'm not suggesting that they should be left! I am thankful
that you want to raise the quality of squeak, I'm sure everyone does.
But, you are taking an active role. Thanks.

Forgive me for missing your point and please bare with me while I ask
another question. There's a difference between tests broken and the code
broken. You said:

"Leaving broken tests in the image does not cause the tests to be fixed.
 Broken tests have been in the image for years and they did not help."

So, you are only talking about removing tests that don't work correctly?
Not tests that work correctly and fail because the code is buggy.

Ok, that makes sense.



-- 
brad fuller
www.bradfuller.com