On Tue, 25 Sep 2018, Jakub Jelinek wrote: > On Tue, Sep 25, 2018 at 05:47:49PM +0200, Martin Jambor wrote: > > > Isn't that bail out too early? > > > I mean most of the warnings that are emitted by the function don't really > > > need TYPE_ARG_TYPES, only the last one does, so can't you just bail out > > > before the last warning? > > > > my reasoning was that if the function is not what I expect it to be, it > > is better not to touch it. On the other hand, I have no problems moving > > this test lower as done in the patch below. > > I guess the question is if we then treat it as a builtin or don't. > Anyway, I'd like to defer that decision to the C FE maintainers.
This patch is OK. (If there's something odd about the argument meaning it ends up not being handled as a built-in, warn_for_abs will have returned early anyway.) -- Joseph S. Myers jos...@codesourcery.com
