JC>On (24 Apr 97) Herman Schonfeld wrote to Carey Bloodworth...
 CB>Let's work through a simple project.  Let's say that we need to
 CB>copy some data from one array to another.  Basically, just like
 CB>memcpy()
 HS> I don't need to be tutured!
JC>Your messages indicate a need for tutoring in both programming and
JC>spelling.
 CB>void memcpy(char to[], char from[], int len)
 CB>{ while (len--) to[len]=from[len];}
 HS> Sloppy code!
 HS> And you claim that optimization creates sloppy code.
JC>I not only claim, but have _proven_ that what you claimed was
JC>optimization produced extremely sloppy code that was considerably larger
JC>and slower than the readable code you started with.
You are lying again Jerry! You lied when you said 'I read the thread' you 
lied when you said 'you need tutoring' and you lied when you said 'i proved'. 
Your constant lies are starting to bother me Jerry.
DJGPP compiles different to MS C, and considering the fact the my message 
was/is/and always will be concerning DJGPP (not MS C, thats right 'NOT') I 
can only conclude to the fact that your results are nothing but invalid.
 HS> First off all, try unrolling the loop, and use pointer
JC>refferencing.
JC>And be prepared for the code to be slower after you do so.  If you have
JC>a compiler that can't unroll loops on its own, throw it away a get a
JC>decent one instead.  My guess is that the only compilers around that
JC>wont' do this on their own are stripped down versions with all or most
JC>optimizations intentionally removed.
JC>    Later,
JC>    Jerry.
... Friday the 13th XXIV: Jason take the Enterprise.
--- Ezycom V1.48g0 01fd016b
---------------