Re: Let's use gcc-4.2, not 4.1 -- OpenMP

Peter Jeremy wrote:
> On Thu, 2006-Dec-14 16:35:41 +0100, Ivan Voras wrote:
>> For what it's worth: +1. It's going to be practically required even for
>> medium-performance applications as CPU clock rate stagnate and more
>> cores are grown.
> 
> Sun have stated that they do not expect SPARC clock speeds to increase
> significantly.  Instead, they will be doubling the number of threads
> per chip every year or so.  (32 now, 64 next year).
> 
> Intel have announced quad-core x86 chips.  AMD will presumably follow suit.
> 
> Based on the current release engineering guidelines, FreeBSD 7.x will
> probably be supported until around 2011 (-RELEASE next year, -STABLE
> for about 18 months, continued support for 2 years after 8-RELEASE).
> The system toolchain is a critical piece of infrastructure and making
> a major change within a release is impractical.  IMHO, rather than
> looking at what is mature now, FreeBSD -CURRENT should be looking at a
> toolchain that is closer to the leading edge (as long as there are
> no significant regressions in stability) but will mature shortly
> before 7-RELEASE.

This is a valid point.

> This maximises the period of "vendor" support for
> the toolchain and therefore reduces the amount of FreeBSD developer
> effort that will be spent supporting the toolchain once the vendor
> stops doing so.

"Vendor support" from the FSF is a myth.  FreeBSD has been chasing this
myth for years, and it never ever turns out to be true.  GCC 3.4 was
rushed in a fast as possible on the exact argument that 3.4 had vendor
support and 3.3 did not.  That was 18 months ago (May 18, 2005), which
is not very long.  In that 18 months, the FSF apparently has stopped
supporting 3.4, 4.0, and 4.1.  Calling that "vendor support" is insane.
The FSF only "supports" the latest and greatest and possibly the
previous release for a short period of time.

> 
> The computing industry moves relatively fast and you can't wait for
> your supplier's product to be fully mature before you start developing
> on it or your customers will go elsewhere because their customers
> won't buy what they see as a product running on an obsolete base.
> 

Yes, the industry moves fast, but that's no reason to fool ourselves
into thinking that the FSF will support GCC 4.2 a day after they release
4.3 and start working on 4.4.  Your point above about the lifespan of
FreeBSD 7.x is a valid one, and I agree that it should be a
consideration.  Vendor support is a myth and should not be a
consideration.

Scott