Re: RS/6000 Optimizer breaks code -- suggestions?

worley@compass.com (Dale Worley)
Wed, 29 Aug 90 09:55:05 EDT

          From comp.compilers
Related articles
RS/6000 Optimizer breaks code -- suggestions? banshee@ucscb.ucsc.edu (1990-08-18)
RS/6000 Optimizer breaks code -- suggestions? jar@florida.eng.ileaf.com (1990-08-21)
Re: RS/6000 Optimizer breaks code -- suggestions? worley@compass.com (1990-08-24)
Re: RS/6000 Optimizer breaks code -- suggestions? daveg@near.cs.caltech.edu (1990-08-25)
Re: RS/6000 Optimizer breaks code -- suggestions? johnv@metaware.uucp (1990-08-27)
Re: RS/6000 Optimizer breaks code -- suggestions? johnv@metaware.com (1990-08-27)
Re: RS/6000 Optimizer breaks code -- suggestions? worley@compass.com (1990-08-29)
| List of all articles for this month |
Newsgroups: comp.compilers
From: worley@compass.com (Dale Worley)
In-Reply-To: johnv@metaware.com's message of 27 Aug 90 20:48:40 GMT
Keywords: C, optimize, debug
Organization: Compilers Central
Date: Wed, 29 Aug 90 09:55:05 EDT

      From: johnv@metaware.com (John Vinopal)


      The union in this case was in fact used as a type cast of sorts; a method
      to prevent a short from overflowing. Bad style and not portable begins to
      sum up this approach. However...


Yes, since the results of using a union for type punning are
undefined (ANSI 3.3, 3.5.2.1), one might say that the code is bad
style. One might also say that it is incorrect, and that the user
can't expect *anything* from it.


Dale Worley Compass, Inc. worley@compass.com
--

Post a followup to this message

Return to the comp.compilers page.
Search the comp.compilers archives again.