Re: Re: How is it possible ???

On Friday 07 January 2005 03:09, Paul de Vrieze wrote:
> While I certainly won't stop you from doing this, please keep in mind that
> there might be rather subtle bugs (depending on your cflags) that occur.
> Also the current gcc streams (every release gets better) can and will
> produce broken code for certain uncommon code/flag combinations. For
> example sse2 support is known to be an issue esp. with the older versions,
> but the newer versions also aren't perfect. As people use different
> compiler versions on different architectures (i386 is the best tested/most
> bugfree), many ebuilds need to be a bit conservative and filter out more
> common flags that are broken for this package.

Actually SSE2 has been one of the problem spots I've had and has been 
responsible for about 4 packages not working properly.  I also use cvs gcc 
builds and have noted a pretty steady decrease in problems over the last 
year.  I wonder if it's worth while adding a feature flag to portage to 
inhibit strip-compiler-flags.  It's too bad that there isn't an easy way to 
do what the Linux kernel devs have done with tainting and have a flag that 
gets included in bugreports indicating you have something like that set.


-- 
Experience has taught me that you feel better on a flight if you avoid
chicken fat in plastic sauce.
        -- The joys of travelling the world by plane
           (Terry Pratchett, alt.fan.pratchett)
...
Patrick Audley                          [email protected]
Blackcat Systems                        http://blackcat.ca
             Bringing Elegance to Complexity


--
[email protected] mailing list