------- Comment #4 from joseph at codesourcery dot com 2008-02-17 23:12 ------- Subject: Re: [4.1/4.2/4.3 Regression] missed evaluation of VM array type when used as a cast
My view remains as I said in <http://gcc.gnu.org/ml/gcc-patches/2006-09/msg00039.html>: front ends should make the internal variable for VLA sizes explicit. This is not however suitable for implementing in development stage 3, let alone in regression-only mode or on release branches. There may of course be a simpler fix for this particular bug. I don't think side effects in VLA size expressions (let alone in VLA size expressions in casts) are that likely to be used in real code outside of conformance testsuites, so I wouldn't consider this regression (worked with 2.95.3, failed with 3.0.4) of great importance. -- http://gcc.gnu.org/bugzilla/show_bug.cgi?id=35198
