On 18.04.2010, at 23:30, Bert Freudenberg wrote:
> 
> On 18.04.2010, at 22:44, Levente Uzonyi wrote:
>> 
>> 
>> 
>> On Sun, 18 Apr 2010, commits at source.squeak.org wrote:
>> 
>>> Bert Freudenberg uploaded a new version of Monticello to project The Trunk:
>>> http://source.squeak.org/trunk/Monticello-bf.388.mcz
>>> 
>>> ==================== Summary ====================
>>> 
>>> Name: Monticello-bf.388
>>> Author: bf
>>> Time: 18 April 2010, 7:11:28.649 pm
>>> UUID: 22c87ccc-e851-4d07-a3e8-a5413d3006b7
>>> Ancestors: Monticello-bp.387
>>> 
>>> - bullet-proof #infoFromDictionary:cache: and re-use it in #extractInfoFrom:
>> 
>> There's a new bug here. MCInitializationTest >> #testWorkingCopy raises an error, because (aDictionary at: #id) returns an UUID instead of a String and UUID class >> #fromString: is not happy about it.
>> Removing the #fromString: send results in 2 errors and a failure.
>> 
>> Probably an #isString check is enough, but it would be better to unify the senders.
>> 
>> 
>> Levente
> 
> 
> Interesting. Why is there a UUID in the dictionary? Ah, I see. Will commit a fix that ensures that the dictionary contains a String. Makes all MC tests green again :)
> 
> - Bert -

Even more interesting, the Monticello tests are green only the first time you run them. I tested in a clean 4.1 image - first time green, second time yellow. The first run leaves behind an MCWorkingCopy named "Monticellomocks". Later it looks for "MonticelloMocks" but can't find it because of the upper-case M ... 

Moreover, the Monticello-Mocks class category is not removed after the tests, as well as the #mockClassExtension method in MCSnapshotTest's "*monticello-mocks" category. This test method actually made it into the Monticello package, so now you can't load Monticello without loading the Test first ... o_O

Also, running the MCTests modifies the "Tests" package. Not nice.

Alas, I found no quick fix.

- Bert -