Wait, what? The second time you run, you get less tests?

That's completely counter intuitive. Tests should never affect future runs... seems like someone left off a tearDown method somewhere. 

On Sep 22, 2010, at 6:53 AM, Levente Uzonyi <leves at elte.hu> wrote:

> On Wed, 22 Sep 2010, Frank Shearar wrote:
> 
>> On 2010/09/22 10:53, Frank Shearar wrote:
>>> On 2010/09/22 09:31, Frank Shearar wrote:
>>>> On 2010/09/22 05:31, Casey Ransberger wrote:
>>>>> Grab it here: http://ftp.squeak.org/trunk/Squeak4.2-10548-alpha.zip
>> 
>> OK, new experiment:
>> * Start up the image,
>> * Run all tests,
>> * Run all tests _again_.
>> 
>> The failing Decompiler tests now all pass.
> 
> No they don't. They just don't run, because those are LongTestCases and there's a test, which turns off the preference that enables the running of LongTestCases. So after the first run, the preference is off, therefore LongTestCases won't be run the second time.
> 
> 
> Levente
> 
>> 
>> frank
>> 
>> 
>