Dear Thomas,
> after some time with a defective computer, I am back online. It seems to be catching.... both my linux laptop and my desktop are as dead as door-nails. > > Here is a fix for PR 52893 (which I just submitted, I wanted to > set a new record between bug posting and patch submissin :-), a > wrong-code regression for trunk and 4.7. Regression-tested. > > OK for both? OK for trunk and for 4.7. As a matter of curiosity, why did you not inhibit common function extraction when the function arguments contain an implicit loop variable, rather than when they are in an implicit loop? That would make the optimisation rather less conservative. Thanks for the patch Paul
