Comments on: A Guide to Undefined Behavior in C and C++, Part 3 http://blog.regehr.org/archives/232 Sat, 18 May 2013 15:30:07 +0000 hourly 1 http://wordpress.org/?v=3.5.1 By: Margit http://blog.regehr.org/archives/232/comment-page-1#comment-2147 Margit Tue, 15 Feb 2011 20:53:10 +0000 http://blog.regehr.org/?p=232#comment-2147 Indeed, and I concur with your explanation.
Indeed it is nasty, and I do not believe that
it is possible (within the C standard) to
code this possibility.
However my code stands and is used universally.

Margit

]]> By: regehr http://blog.regehr.org/archives/232/comment-page-1#comment-2145 regehr Tue, 15 Feb 2011 20:09:15 +0000 http://blog.regehr.org/?p=232#comment-2145 Hi Margit,

The ANSI C (C89) standard says: “The end of a full expression is a sequence point.”

Anyway, the real reason I’m not that interested in pursuing this further is that the developers of several major C implementations have decided what to do. It’s fine if you believe this is a bug, but I strongly doubt they would agree. The point of my post was to call this issue to the attention of C developers, since it’s nasty.

]]> By: Margit http://blog.regehr.org/archives/232/comment-page-1#comment-2144 Margit Tue, 15 Feb 2011 19:48:51 +0000 http://blog.regehr.org/?p=232#comment-2144 Sorry to hear that. I was hoping to include my
observartions in your post.

To iterate, a may NOT reorder instructions when
a sequence-point occurs.

To my mind, your obervations for Clang/Intel represent a bug.

Margit

]]> By: regehr http://blog.regehr.org/archives/232/comment-page-1#comment-2143 regehr Tue, 15 Feb 2011 19:31:25 +0000 http://blog.regehr.org/?p=232#comment-2143 Hi Margit, I’m losing interest in this thread.

]]> By: Margit http://blog.regehr.org/archives/232/comment-page-1#comment-2142 Margit Tue, 15 Feb 2011 19:29:20 +0000 http://blog.regehr.org/?p=232#comment-2142 But we are not talking about C99!

Margit

]]>