[Chandler-dev] Unit and Functional Tests updated to handle Unicode character display

[Markku Mielityinen]

I will have to agree with Heikki and Philippe, see my original comments 
from Tuesday (Bug 5952). Having this kind of randomness is dangerous and 
instead maybe we should run the test for each of the characters but 
using only one character at a time. This is my 2 cents on the matter...
> Heikki Toivonen wrote:
>> That sucks for a developer. Instead of running the test once, you need
>> to run it n number of times to get y percentage confidence that your
>> code is working. You check in, and you might get a failure immediately,
>> an hour later, or a day later or, ...
>>   
> Agree 100%. Also remember that we ping the devs who committed code 
> between a green build and a failure. If we start having random 
> failures, people will start to be complacent again and failures won't 
> be fixed in a timely fashion.
>> The problem with TestCrypto was not determinism, the problem was that it
>> used only one code point with value that fit into "extended ASCII".
>>
>> My suggestion is to select two (or if that is too few, a couple more)
>> representative code points and put them always in the string. So
>> something like:
>>
>> u = u"%s%s%s"% (u'\u00C5', defaultText, u'\u8ad6')
>>   
> +1
>
> - Philippe
> _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
>
> Open Source Applications Foundation "chandler-dev" mailing list
> http://lists.osafoundation.org/mailman/listinfo/chandler-dev