On 08/26/2013 12:46 AM, Joseph Rushton Wakeling wrote:
On 26/08/13 00:36, Timon Gehr wrote:
This is uncalled for.
I'm sure you didn't mean it, but your original remark, "Why would that
be the case?", came across as flip and dismissive about a serious
practical problem
It was a genuine question. I assumed that the inliner actually works.
But you are right, DMD does not even seem to inline iota, and gdc
appears to fail to vectorize some iota-loops even though it vectorizes
the range-foreach equivalent.
