Hi! Seems we already handle delete expressions the way the DR clarifies, so this patch just adds a testcase which verifies that.
Tested on x86_64-linux, ok for trunk? 2024-09-25 Jakub Jelinek <ja...@redhat.com> * g++.dg/DRs/dr2728.C: New test. --- gcc/testsuite/g++.dg/DRs/dr2728.C.jj 2024-09-25 13:44:15.102534622 +0200 +++ gcc/testsuite/g++.dg/DRs/dr2728.C 2024-09-25 13:44:09.876605940 +0200 @@ -0,0 +1,20 @@ +// DR 2728 - Evaluation of conversions in a delete-expression +// { dg-do run } + +struct S { + S (int *x) : p (x) {} + operator int * () const { ++s; return p; } + int *p; + static int s; +}; +int S::s; + +int +main () +{ + int *a = new int; + S s (a); + delete s; + if (S::s != 1) + __builtin_abort (); +} Jakub