> His point, I think, was that i386-redhat-linux is quite a bit more
> accurate and useful than i386-gnu-linux. Ditto for i386-suse-linux
> and i386-debian-linux.
Is it? The same package ought to work on all of them, and if it doesn't,
you _should_ (you won't always) be able to tell from the .deb or .rpm
extension, not from the name of the package.
The main reason for this? They all use glibc2.
How 'bout we call it 1386-glibc2-linux.
Oh. Damn. I just remembered what the g in glibc stands for.
Obviously we can't have that then. ;)
> But to take {redhat,suse,slackware,debian} and call them all gnu is
> (a) incorrect,
It is? Which one of these does not use GNU-owned software for most of it's
basic userspace components? GNU certainly does not have a monopoly on
such, but it certainly is responsible for more than any of the
distributions. It seems the best approximation to correctness.
> (b) somewhat annoying,
To you. To others it's annoying the other way. Such is life.
> (c) loses important information.
If the software won't run on all platforms that could be called
i386-glibc-linux, then the fault lies in the software.
If it is inherently distro-specific, then you don't need to have the
platform as part of the name.
Anyone that wants to debate the gnomenclature issue, should be aware
that it is going to take an extended discussion that you should _not_
invite here. If it was possible to demonstrate either side of the argument
in the space allowable to an offtopic thread in a technical mailing list,
then the issue would have died long ago.
Having said that, I hope the following might help people see that the
topic is worthy of it's own discussion area, and that you cannot trivially
brush Richard&co. off, however much you would like to end this tiresome
thread.
GNU own a very small proportion of the code we run, on the whole, but what
they do own comprises a great many of the most fundamental and universal
components of a system. Depending on your definition of "operating system"
GNU might own a very large proportion of it, or a minor part if you count
everything on your distro CD, or none if you equate "OS" with "kernel".
When you are considering a "platform" for porting purposes or similar,
what matters most (after the CPU) is the standard library. If you port to
glibc, you can create a single i386 binary that can run on Linux, FreeBSD
and Hurd, but _won't_ run on libc5-based Linux.
I know a lot of people won't agree, but hopefully they can see that there
is enough depth here that the issue will never be laid to rest within the
narrow confines of this list.
-Greg Mildenhall
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/