------- Additional Comments From gdr at integrable-solutions dot net 2005-07-24 03:21 ------- Subject: Re: partial_sum is too constrained
"bangerth at dealii dot org" <[EMAIL PROTECTED]> writes: | No, the standard says that the result for an iterator 'i' in the output | range is | ((...(*first + *(first + 1)) + ...) + *(first + (i - result))) | So arithmetic is done with the data type of the input range. However, there is a snag. In the example submitted by the reporter, "first" is of type "char*". So any arithmetic happening, starts by *promoting* char to int before computation is done. Consequently, it appears legitimate to expect that intermediate result being used throughout the accumulation, because it is the natural type of the result. | What |Removed |Added | ---------------------------------------------------------------------------- | Status|UNCONFIRMED |RESOLVED | Resolution| |INVALID I don't think it is invalid PR. We may, however, eithed do The Right Thing; or submit a DR; or do both. -- Gaby -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=22634
