Hakan Kuecuekyilmaz wrote:
> On Do, 2008-07-17 at 17:33 +0300, Andrey Hristov wrote:
>> Hakan Kuecuekyilmaz wrote:
>>> On Do, 2008-07-17 at 15:51 +0200, Vladislav Vaintroub wrote:
>>>>> -----Original Message-----
>>>>> From: Hakan Kuecuekyilmaz [mailto:hakan@stripped]
>>>>> Sent: Thursday, July 17, 2008 3:44 PM
>>>>> To: Vladislav Vaintroub
>>>>> Cc: commits@stripped
>>>>> Subject: RE: bzr commit into mysql-6.0-falcon branch (hakan:2750)
>>>>>
>>>>> On Do, 2008-07-17 at 14:37 +0200, Vladislav Vaintroub wrote:
>>>>>> Why 't1' ? Is 't' not good enough?
>>>>> t1 is standard name for test tables like t1, t2, t3, t4, ..., tn
>>>> Please send the link to standardization committee page:)
>>> It is documented here
>>> http://dev.mysql.com/doc/mysqltest/en/writing-tests-naming-conventions.html
>> sometimes the documentation is outdated and what the devs do is always
>> one-step ahead.
>> If you can explain me why sporadic failure will increase if table names
>> are unique? Because things are broken now and it will surface when you
>> do it the new way?
>
> Sorry, I can't follow you here. How things should be broken when all
> existing tests are meant to have t1, t2, t3, ..., tn as table names?
>
> Because you will not detect side effects in new tests, if you use unique
> table names. But at one point those side effects will kick back in.
> Hence my example with IE rendering complete broken HTML into something
> and giving the web-developer the illusion of something correct.
False negatives? Some table wasn't deleted, next test start to test, you
need to check why the next test fails. Next test uses t1 like the
failing test has similar or the same structure, just about different data.
>
> Cheers und Küsschen
>
> Hakan
Best,
Andrey
