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

Re: ok, now for my paranoid ranting

In article <199806301600.MAA00344@mike911.clark.net> you wrote:

: why not simply having Debian GNU/Linux and Debian GNU/HURD, as I'm sure has 
: been proposed already? the same os with different kernels and other less 
: subtle things, such as the filesystem hierarchy.

I've studed the Hurd a bit, and sat through MB's tutorial at the first (and 
so far only?) FSF conference a couple of years ago.  Anyone who is worried 
about global domination by the HURD anytime soon should calm down...

>From a purely theoretical perspective, the fact that the HURD won't share the
Linux syscall interface, and is diverging on filesystem layout, means that
the free software community at large is going to be duplicating a lot of work
on applications.  The syscall interface could be handled through glibc fairly
cleanly, but the file system layout essentially guarantees that anything that
needs to run on a HURD system will have to be recompiled from sources.  Given
the amount of manpower it takes to build each package once for each processor
architecture for Debian, adding another level to the package-building nested 
loop pains me.  C'est la vie.

On the other hand, the HURD team are obviously convinced that being compatible
on that sort of level with anything that currently exists is not important.
They clearly believe that what they are building will be "better".  The 
performance issue means that if they built the HURD in such a way that you 
could load a system with Linux and the HURD and a bunch of applications and
run either kernel environment... then the HURD would just plain lose, since
real people mostly only care about their applications, and they want them to 
go fast.  

Thus, I believe that the approach the HURD community is taking is the only 
one that they *can* take.  If their ideas really are better, we'll all know 
it when they are done, and more power to them!  I agree that it will be 
*quite* different, but it's going to take convincing over time for me to 
believe that the things they propose are actually "better".  

Given this reality, the level on which it makes sense to cooperate is pretty 
obvious to me...  Anything that helps the free software community get better 
installation tools is good.  Having another highly visible project using dpkg 
and the .deb package format is good.  I'm sure we'll find other similarly 
scaled places to cooperate that will be good... improving the tools for
automatic package recompilation springs to mind.

So, bottom line, working with the HURD guys is good... as long as we don't
lose sight of what we're doing with Debian in the meantime.


To UNSUBSCRIBE, email to debian-devel-request@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org

Reply to: