Conversation

Notices

1. Embed this notice
   aks (sqx@mastodon.social)'s status on Sunday, 10-Nov-2024 06:54:13 JST aks

   @vampiress I stubbed my toe on a text book variant of this very recently where I simply had to duplicate the top stack element, like:

   stack[stack_height++] = stack[stack_height-1];

   The RHS was "hidden" behind a function call which made it slightly harder to see straight away. I didn't see the bug until I tried compiling it on clang instead of gcc.

   So.. I'm half-seriously looking into Ada now :-) (also because it's not the only recent disappointment I've had with C)

   • Embed this notice
     pinskia (pinskia@hachyderm.io)'s status on Sunday, 10-Nov-2024 06:54:12 JST pinskia

     in reply to

     @sqx @vampiress At least GCC does warn:
     <source>: In function 'f':
     <source>:3:21: warning: operation on 'stack_height' may be undefined [-Wsequence-point]
     3 | stack[stack_height++] = stack[stack_height-1];
     | ~~~~~~~~~~~~^~

     clang warns too.
     I almost always use `-W -Wall` (-W is an alias for -Wextra) when compiling which is why I see these warnings.
     Also there is a paper for C to make it well defined and not unspecified.

   • Embed this notice
     pinskia (pinskia@hachyderm.io)'s status on Sunday, 10-Nov-2024 07:25:30 JST pinskia

     in reply to

     @sqx yes that is definitely much harder to warn about. I can't find the paper which is written up to change the standard. The paper is also requesting it be considered a defect against older versions of the standard. Really I wish this was not unspecified but since it is GCC tries to takes advantages of it. Newer C++ makes some things not unspecified but still leaves out some stuff like arguments evaluation.

   • Embed this notice
     aks (sqx@mastodon.social)'s status on Sunday, 10-Nov-2024 07:25:31 JST aks

     in reply to

     • pinskia

     @pinskia (NOTE: I untagged Elissa because she has since deleted the thread). I get the same result with that exact code, but in my case the RHS was inside a function call, and then both gcc and clang are silent about it, even with -fsanitize=undefined. So the code in the image doesn't warn me in gcc/clang, but if you change #if 0 to #if 1 it does