tech-toolchain archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

Re: Stop shipping static libraries for NetBSD




On 18-Aug-08, at 6:11 PM, Matt Thomas wrote:

The proposal is stop shipping; not stop using. Anything in the build can continue to use them, as can anyone doing their own builds. The proposal is to not ship them so the average user won't be able to link static programs.

To me it is just as sad that NetBSD, in its role as as a binary-only vendor, would like to take the same attitude that Sun and Apple and Microsoft have done and also force everyone providing source-based software to those using those binary-only distributions to be totally locked into the "everything must be dynamic all the time" mindset.

It just doesn't make any sense whatsoever no matter which way I look at it.

Of course I just don't get the whole mentality of using binary-only distributions of open-source systems in the first place.

In any case one can choose to lock one's self into requiring dynamic runtime linking all the time even when sane alternatives exist and regardless of whether one has source or not, but I think doing so is rather short-sighted and wasteful in the worst of ways.

Why not just crunchgen the whole system into one "lean" binary and maybe even lock it right into the kernel? Then you get the benefits of being able to upgrade your whole system just by booting a new kernel! No runtime overhead, and not one byte of waste through duplication of object code either. Pure and total reuse everywhere all the time but never any possibility of DLL hell! What an amazing idea! Come on now, let's get mozilla into the kernel! And it'll always just be one program to recompile. It just can't get any simpler!

Or why not just run a fully interpreted system, or at least put a just- in-time compiler into the kernel -- much more robust and even more dynamic. Never recompile again! Resurrect that machine independent, compiler intermediate object format from the OSF days for the proprietary vendors. Performance? No worries since everyone has multiple multi-GHz cores in every system they could ever want to use. Security? Well we can put that in later once the JVM folks finally figure out how to do it properly.

Can someone involved in this crazy proposal _please_ go re-read the published goals for NetBSD?

--
                                        Greg A. Woods; Planix, Inc.
                                        <woods%planix.ca@localhost>



Home | Main Index | Thread Index | Old Index