gcc 3.4 vs 4.0

Ed Hill ed at eh3.com
Thu Nov 24 13:38:39 EST 2005


On Thu, 2005-11-24 at 09:38 -0500, Stephen Adler wrote:
> Thanks for the reply. Since I'm actually writing the software, or 
> building it from source, I'm in the position of  defining what target 
> platform to do the development on. Thus if I choose gcc 4.0, it would be 
> because it would have better optimizations, etc, but the down side would 
> be that its buggier. Some people will never adopt a X.0 version of a 
> package since they deam it unstable, (and I know for a fact this was the 
> case with 4.0.0, my matrox G450 didn't work after I did an upgrade to 
> FC4, it took about 6 months for a patched version of gcc to come out and 
> the associated xorg updates.) Anyway, any help or suggestions of the 
> merits of gcc 4 vs gcc 3 is greatly appreciated.

Hi Stephen,

It really shouldn't be an either/or choice.  Since you have the source
code, you can easily install copies of both GCC 3.x and 4.x on your
development box and build/test/etc your application with both of them.  

And in terms of gcc 4.x vs. 3.x, I've had little in the way of problems.
GCC 4.x is more strict about syntax and it actually *helped* point out a
few non-standard idioms in our code.  But to a very great extent, code
that works well with 3.x will also work nicely with 4.x.  And, as you
pointed out, you might even see some run-time performance boosts.

Ed

ps - The example you provided with the Matrox driver and GCC 4.x 
     is kernel code and thats a whole different ball of wax.  If 
     you're writing user-space code and have all the source, then 
     you have a *LOT* fewer things to worry about -- many distros 
     are already relying on GCC 4.x and encountering few problems.

-- 
Edward H. Hill III, PhD
office:  MIT Dept. of EAPS;  Rm 54-1424;  77 Massachusetts Ave.
             Cambridge, MA 02139-4307
emails:  eh3 at mit.edu                ed at eh3.com
URLs:    http://web.mit.edu/eh3/    http://eh3.com/
phone:   617-253-0098
fax:     617-253-4464




More information about the Discuss mailing list