[Info-vax] BSDs vs Linux (was Re: New CEO of VMS Software)

[Dan Cross]

In article <uo60k6$1f897$1 at dont-email.me>,
Simon Clubley  <clubley at remove_me.eisner.decus.org-Earth.UFP> wrote:
>On 2024-01-15, Scott Dorsey <kludge at panix.com> wrote:
>> Lawrence D'Oliveiro  <ldo at nz.invalid> wrote:
>>>Pro tip: If you have to start to resort to personal attacks, that’s an 
>>>admission that your argument has failed.
>>
>> Let me explain to you what Linux is, then.  Linux is the Linux kernel, as
>> signed off on by Linus T.
>
>Linux is generally regarded to be an entire distribution, with the Linux
>kernel as one component. My Android phone runs the Linux kernel, but it
>is not Linux.

This is not really true.  Linux is really just the kernel; that
is why people make such a stink about "GNU/Linux" when referring
to Linux distributions (though there are such that don't use the
GNU tools, like Alpine Linux).

>> You can take the linux kernel and modify it for your own purposes, yes.
>> Since it's kind of a mess and not really very modular, this turns out to
>> be a difficult thing to do, but it's a thing that is possible to do.
>> It's a good bit harder than modifying the BSD kernel.
>
>It is _vastly_ more modular than the VMS kernel. Unlike the VMS kernel,
>I can add my own filesystems for example, and they will just integrate
>into the rest of Linux. In fact, I can add that new filesystem as a kernel
>module, so I don't even need to touch the Linux source code. Try doing
>that on VMS. :-)

That may be the case, but the comparison was to the BSD kernel,
not to VMS.

However, the whole supposition on the troll's part ("there's
just one kernel in a thriving ecosystem of distributions!  Why
can't those idiots over in the BSD world do THAT?!") is based on
a false assumption.  The reality is that big organizations that
use Linux almost always fork the kernel: look at any of the
hyperscalers, or any number of OEMs of consumer electronic
devies, for example.  There are 1000s of versions of the "Linux
kernel".

>> But once you have done it, it's not Linux any more.  It is something else.
>> You can call it what you want, but you can no longer call it Linux.  Not
>> unless you can get Linus to sign off on your changes.
>
>That is misleading. If you modify the Linux scheduler (for example), you
>have a point. If you add a new filesystem (for example), then it most
>certainly is still Linux.

I think the point is, if you fork the kernel for whatever reason
(and many times that's not because one _wants_ to but because
one _needs_ to for some reason, such as one can't afford the
time or doesn't have the political capital to get a critical
patch upstreamed, so they end up floating it themselves), then
you've got a system that's Linux-derived, but you're not keeping
up with what "Linux" is in the rest of the world.

That's a huge cost to bear.

>> All of the "lack of fragmentation" that you think is so wonderful about
>> Linux is exclusively the consequence of this.  The fact that one person
>> controls what is and what is not the Linux kernel is why there is just
>> one Linux kernel family and why there is no fragmentation as there is
>> with BSD.
>>
>> This is why your talk about fragmentation and asking why BSD versions
>> are not all the same makes you look like a troll.  Because there is
>> something very specific going on with Linux that prevents that 
>> fragmentation.  Some people like that.  Not everybody does.
>
>Lawrence is coming across as a Linux zealot, who is out of touch in a
>number of areas. However, when he is not making some crazy out of touch
>comments, such as replicating the VMS APIs on Linux work, then he is also
>making _some_ valid points.

His problem is that he doesn't seem capable of recognizing where
his assumptions break down, so he just repeats the same thing
over and over, regardless of evidence to the contrary.  The
Linux ecosystem is actually _enormously_ fragmented; he just
doesn't know about it, so assumes it's not.

	- Dan C.