incubator-stdcxx-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Liviu Nicoara <nico...@roguewave.com>
Subject Re: 23.containers.deque.modifiers.cpp
Date Wed, 25 Jan 2006 21:05:34 GMT
Take three: I have attached the two deque tests (modifiers and special).
I have replaced the ToString class with the special formatter X, added
special options to disable the different types of iterators exercised
(the "right thing" too).

I still see the bogus "leaked 0 objects". I could look into it next.

Liviu


Liviu Nicoara wrote:
> Martin Sebor wrote:
> 
>>Liviu Nicoara wrote:
>>
>>>Take two with corrections to add z modifier to size_t arguments.
>>
>>I'm afraid that's still slightly incorrect and, in fact, has undefined
>>semantics according to 7.19.6.1, p9 of C99. size_t is an unsigned type,
>>and the set of format specifiers that match an unsigned type is o, u,
>>x, or X, but not d. This is easy to fix, though.
> 
> 
> My bad, I verified only against POSIX and sed-ed the previous macro
> (_RWSTD_PRIz).
> 
> 
>>What may be more tricky is fixing the formatting of the X class -- I
>>see some strange/incorrect formatting when tracing the execution of
>>the test with --trace:
>>
>># TEXT: line 298: deque((null)).assign(0, 65): leaked 0 objects after a 
>>successful insertion
> 
> 
> I'll check if it still happens after the %X change.
> 
> 
>>Also, it would be nice to add command line options analogous to those
>>we've implemented for the algorithms tests (i.e., those controlling
>>which functions and/or which of their specializations to disable).
> 
> 
> I'll add them.
> 
> 
>>Finally, since you split up the original test into two (and made
>>the part that exercises swap a standalone test) I would like to
>>commit both at the same time to make sure we don't forget one.
> 
> 
> Will do.
> 
> Thanks,
> Liviu
> 


Mime
View raw message