Subj : Re: OO compilers and efficiency
To   : comp.programming
From : Chris Dollin
Date : Wed Jul 20 2005 11:49 am

Rob Thorpe wrote:

> Chris Dollin wrote:
>> Brian wrote:
>> > Anyway, those are some of the things that kind of nag at me.
>> > I think it's a safe statement to say C can beat any OO compiled
>> > program pound for pound given the same programmer skill and
>> > adherence to language goals.
>>
>> There are so many assumptions buried there that I can only
>> mention a few.
>>
>> * OO languages typically [ie, Java] come with GC. C doesn't.
>>   This can make an enormous difference to the simplicity of OO
>>   code. That means that OO programs can tackle larger problems
>>   with the same-ish amount of engineering effort  - they can
>>   out-produce the C programmer before the code efficiency is
>>   an issue. (What the OO program loses in GC costs is roughly
>>   the same as what the C program loses in memory-management
>>   costs, eg copying.)
> 
> I think your other reasons are stronger than this one.  Although C
> doesn't come with a GC it's very simple to add one, you can download
> the Boehm garbage collector and use it for example.

If the C program wasn't written with GC in mind, randomly using
the Boehm GC (and tweaking malloc/free) won't get you the advantages 
of a GCd language. 

If the C *was* written with GC in mind, then you're writing in a
C-like GCd language with the usual benefits - assuming that you
don't fall over any of the conservative GC assumtions. But that
language isn't really C any more.

> All the questions of efficiency depend on how important efficiency is.
> When it is important you may have to reject a language or paradigm
> because it can't give you the performance you need.  When efficiency
> isn't important you can pick and choose whatever language you think is
> appropriate.

Verily.

-- 
Chris "electric hedgehog" Dollin
It's called *extreme* programming, not *stupid* programming.

.