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

Re: SEUL: RFC: Base SEUL on Debian?



William T Wilson wrote:
> > How is that going to work given that any one eliment of the core can be
> > revised at a different time than the other elements?  You going to hold
> > a bug fix or security fix until the next core version?
> 
> No; the core can be updated if necessary.  We go from version 1.1 to
> 1.1.2, say, if we have some library buffer overflow or something.  Most
> bug fixes and security fixes don't impact functionality, ease of
> compiling, or much of anything.  And with luck, there won't be many bugs
> or security holes in the core.  By the time this is done, glibc will be
> reasonably stable, and the only 'continual source of holes' in the core
> will probably be X-windows.  And it's as yet undecided how much of
> X-Windows will actually be in the core.

But if I were a developer and worked on program "corefoo" and found a
glitch in it after core 1.3 was out, how could I version corefoo?  I'm
assuming you're saying that programs in core are numbered with core i.e.
"corefoo 1.3" in this case.

The alternative is that a README will say it is for core 1.3 even though
it may be corefoo 4.6.2, in which case if corefooplus 4.6.2 needed
corefoo 4.6.2 to work properly but corefoo 4.6.2 didn't need corefooplus
at all, how would you know corefoo 4.6.2 is in core 1.3 instead of
corefoo 4.6.0 which corefooplus can't work with?

If you can make heads or tales of that. LOL.

> > So you're not talking about the core binaries needed for a base system?
> 
> No, we are.  However, core binaries aren't much different from
> distribution to distribution.  It's the application packages that are
> different.

That's what I meant by not being their own dis'.  The vi-war is a
perfect example.  

Example only:
Suppose Slackware thinks that ae is the best editor for a core dis' but
core is established as using vi in the core.  The origin of differing
distributions was from people thinking that their idea of a rounded dis'
is one that uses one set of programs as opposed to another so they went
off and put their own together.  

Granted, I don't want to get rediculous about it, there are a lot of
bare essentials that you just have to have for a working system, but it
sounds to me like you are talking about more than just the baseline
utilities needed for it to boot and have a filesystem.  Or am I wrong
here?  Because I'm sure that there is already a universal absolute
minimum set that everyone uses, there has to be for it to work.

> If RH manages to patent the idea for this core, I'll be first in line to
> take out patents on going to the store and going to bed at night.  

I was really just using RH as an example since, from what I've been
reading, they seemed to be against getting involved.

So let's say RH doesn't but M$ patents the idea (algorythm) of a unified
Linux core structure.  Puts the kibosh on linux becoming universal
enough for vendors to port to doesn't it?

Actually maybe it falls more under a copywrite than a patent.  Who
knows, just a thought.