nanogui: State of the nanogui union?


Previous by date: 29 Sep 1999 13:05:58 -0000 Re: State of the nanogui union?, Alan Cox
Next by date: 29 Sep 1999 13:05:58 -0000 Re: State of the nanogui union?, Bradley D. LaRonde
Previous in thread: 29 Sep 1999 13:05:58 -0000 Re: State of the nanogui union?, Alan Cox
Next in thread: 29 Sep 1999 13:05:58 -0000 Re: State of the nanogui union?, Bradley D. LaRonde

Subject: Re: State of the nanogui union?
From: "Vidar Hokstad" ####@####.####
Date: 29 Sep 1999 13:05:58 -0000
Message-Id: <19990929130112.26980.qmail@mail.relight.com>

On Wed, 29 Sep 1999 08:37:05 -0400 you wrote:
>----- Original Message ----- 
>From: Vidar Hokstad ####@####.#### 
>To: Bradley D. LaRonde ####@####.#### 
>Cc: ####@####.#### 
>Sent: Wednesday, September 29, 1999 8:33 AM 
>Subject: Re: State of the nanogui union? 
> 
> 
>> On Wed, 29 Sep 1999 08:15:41 -0400 you wrote: 
>> >Everyone one: please brace yourselves for this next question.  :-) 
>> > 
>> >How about C++ instead of macro hackery? 
>> 
>> I thought about it, but decided against it for since reasons. For our 
>> box it isn't that critical, since we'll have the C++ libraries on it 
>> anyways, but for PDA's etc., where including a C++ library might consume 
>> almost all of the available memory, whereas with C you could get away 
>> with only a very stripped C library, C++ is a nuisance. 
> 
>So... strip the C++ library.  I'm not saying STL, I'm saying classes. 

Doesn't help.... You'll still have the C library around, so going C++
means that you'll add yet another (quite big) library too, as opposed to
being able to get by with just a C library, maybe even with quite a bit
of functionality taken out.

Also, many embedded platforms and PDAs doesn't have a proper C++ library,
or even C++ compilers that can generate code suitable for them.

STL wasn't in my mind at all...

>> Actually, I work quite a bit with C++, but just not for size critical 
>stuff. 
> 
>Isn't that just a myth? 

Not in my experience. I've several times rewritten C++ applications that
have been carefully optimized for performance in pure C and reduced them
to as littles as 20% of the original, with 50% being typical.

One important note here: This is of course C++ applications which
actually are C++, not just C compiled with a C++ compiler - if you don't
use classes or functions from the C++ library size shouldn't be an issue
any longer. But why use C++ then?

It's also with egcs - I don't know if the numbers would be equivalent
with other complers.

In most cases with C++ apps that aren't carefully optimized, the numbers
can be even more dramatic... Few C++ programmers seem to realize how to
use the string class efficiently, for instance - I've seen programs where
more than 50% of the time was spent concatenating and copying strings 
because they didn't realize how string expressions were compiled.

Regards,
Vidar

Previous by date: 29 Sep 1999 13:05:58 -0000 Re: State of the nanogui union?, Alan Cox
Next by date: 29 Sep 1999 13:05:58 -0000 Re: State of the nanogui union?, Bradley D. LaRonde
Previous in thread: 29 Sep 1999 13:05:58 -0000 Re: State of the nanogui union?, Alan Cox
Next in thread: 29 Sep 1999 13:05:58 -0000 Re: State of the nanogui union?, Bradley D. LaRonde


Powered by ezmlm-browse 0.20.