[PATCH] 4.0 fix for libstdc++/23871

Paolo Carlini pcarlini@suse.de
Wed Sep 14 21:16:00 GMT 2005


Paolo Carlini wrote:

>>>I would suggest simply adding the testcases that originally went in
>>>together with those bits of the patch. In other terms,
>>>testsuite/27_io/basic_ostream/inserters_arithmetic/char/7.cc and
>>>testsuite/27_io/basic_ostream/inserters_arithmetic/wchar_t/7.cc
>>>(actually the bits for long and long long should pass already, please
>>>double check).
>>>
>>both of those tests already pass with the 4.0 branch, but the new test
>>fails without the patch and passes with it.
>>
>Janis, this is impossible, sorry. Please check again, carefully: of
>course I added those tests to mainline exactly to test those bits and if
>you care to look closely to the patch and to the test, you will notice
>that the logic error in the code is precisely tested by the testcase. If
>you find a flaw in the test, please tell me ASAP! Only the bits for long
>and long long are already ok, if you want you can remove the second half
>of 7.cc, but the first part is certainly ok as-is (just checked again,
>anyway).
>  
>
Of course, in case you actually meant that *the second half* of 7.cc
already passes with the 4.0 branch, then I misunderstood you, apologies.

Anyway, my point stay: let's do the simplest thing, let's just backport
to 4.0 the patchlet +  its accompanying testcases (char/7.cc,
wchar_t/7.cc) in the *very same* form of mainline.

Patch pre-approved as soon as 4.0 reopens.

Paolo.



More information about the Libstdc++ mailing list