[squeak-dev] Re: Bug in BytecodeGenerator>>#jump:if:
hmm at heeg.de
Sun Jun 22 15:38:08 UTC 2008
> But my joke was about the dogma "provide a failure test before your
> change is accepted in Squeak".
> Correcting such bug is ten times easier than writing a TestCase, and
> is an emergency. So in some exceptional cases like this the dogma
> should be "Correct immediately then provide a non regression test case".
<offtopic>As a christian I'm pretty cautious about dogmata - they are
always formulated by people who have only partial insight and don't know
the whole truth - as Paul wrote in 1 Cor. 13,12... So I take them with a
grain of salt.</offtopic>
If you only accept SUnit tests, you're bound to encounter cases such as
this where it is not possible to write a meaningful failure test.
But when you accept other kinds of tests, it's pretty simple. The test
would be a method which when compiled and executed crashes the image.
With the fix in place, the image does not crash.
The most important feature of tests is that they are repeatable, which
this test would obviously be.
More information about the Squeak-dev