[conspire] Linux also rans

[Rick Moen]

Quoting Texx (texxgadget at gmail.com):

> In the beginning there was debian...  (grin)

You're kidding, right?  Early Linux distributions started with HJ Lu's
boot/root images (which I ran) in 1991, Then MCC Interim Linux (1992),
TAMU Linux (1992), Softlanding Linux System (SLS, 1992), Yggdrasil
Plug-and-Play Linux (1992), then Slackware (1993), then Debian (1993),
then Red Hat Linux (1994).  (I may be forgetting a few, and of course
many came after the ones I listed.)

> Then a bunch of North Carolina "frat boyz" forked a branch off.  To my
> knowledge, RedHat was the first shrink wrapped release.
> 
> Correct me if Im wrong?

Yggdrasil Plug-and-Play Linux was substantially earlier.  Red Hat
Linux's first non-beta was released in 1995 (following a test release
and a beta in late 1994).

> After that the forking went crazy.

It wasn't forking in the usual sense of the word -- though Slackware was
originally based on SLS.  Debian wasn't _based_ on SLS, but inspired by
it in the sense that Ian Murdock found that SLS sucked so much that he
founded Debian out of frustration.

> Probably due to it being first shrink wrapped, RedHat seems to have
> taken a strangle hold in the US markets?

No.  Better marketing and vulture-capital funding, mostly.  For example,
Yggdrasil was (as far as I know) only ever just Adam J. Richter and Bill
Selmeier with an office off Stevens Creek Boulevard.

> Disagreements between RedHat & the userbase and possibly some
> copyright issues brought Centos up to share the stranglehold on the US
> market.

No.

RHEL was derived from the former RH Advanced Server 2.1 product, and is
under trademark-based encumbrances that the RHL distro had lacked.
Quite a number of outside parties realised around 2002(?) that all of
the software contents of the constituent SRPMs remained open source,
though two SRPMs (redhat-logos and anaconda-images) holding
trademark-encumbered image files (but no software) were not, hence it
was feasible to create a 100% unencumbered unbranded derivative just by
replacing the image files and recompiling.  This matter was widely
discussed on a mailing list called 'rhel-rebuild' and gave rise to many,
many RHEL rebuild projects, of which only a few still survive almost two
decades later.

I believe the first was John Morris's White Box Enterprise Linux.
Others included Greg Kurtzner's cAos Linux, Tao Linux, Pie Box
Enterprise Linux, Lineox's two distribution, BioBrew Linux, X/OS Linux:,
StartCom's four distributions, Engineering Computing GNU/Linux (from
China), Eadem Enterprise AS, Fermi Linux, NPACI Rocks Cluster
Distribution, Lawrence Livermore National Labs Linux, Scientific Linux,
Oracle Linux, and CentOS.

Most of those have gone away (it's been a long time, after all).  The
various HPC cluster distributions have mostly converged on Scientific
Linux, because a bunch of high-energy physics labs standardised on it,
but Fermilab keeps Fermi Linux alive as a custom _variant_ of Scientific
Linux (I don't know why), and Rocks Cluster Distribution (they dropped
the 'NPACI') is still around.

At the time I was doing HPC clusters for a living, the HPC distributiosn
weren't just RHEL rebuilds, because it was necessary to add a number of
important packages (such as MPI, Lustre, and Ganglia) that aren't in
RHEL.  I'm not sure that's still the case with Scientific Linux, but it
is with Rocks Cluster Distribution, at least.

> Seeing Ubuntu suddenly pop up in the US commercial market was a
> surprise and when I asked about it duting an interview, I was told
> that RedHat & Centos got sloppy about maintaining their driver bases,
> while Ubuntu was much better on the drivers.  Therefore they were
> taking a kernel performance cut to have a decent driver base.

That seems totally wrong  (And I don't know why you considered it a
'surprise'.  It was market opportunism.)

Ubuntu is a close derivative of Debian, making a market-driven decision
to release for only the three most market-significant CPU architectures
(PowerPC, i386, and AMD64, of which PowerPC has now been dropped, i386
is being dropped, and AAarch64 is coming in), at a time when, IIRC,
Debian fully supported 14 CPU architectures.  Canonical, Ltd. also 
promised for Ubuntu a very reliable, pre-announced release schedule.

And they do relentless marketing, but on the cheap, because Canonical,
Ltd. isn't actually much of a company, which is why it's chartered in
Isle of Man, a tax haven.

Canonical does not do any significant work on drivers, which is in
marked contrast to Red Hat, Inc.  Whoever told you that was, um,
differently informed.

> Suse seems to be barely hanging on in the US.  Is it being displaced
> here by RedHat/Centos?  Or is it tainted by association with Novell ?

You and I are among the few who even think about Novell.  And why on
_earth_ would the former connection 'taint' that distribution?

SUSE just did another change of corporate ownership to EQT Partners, a
Swedish private equity group, after being a unit of British firm Micro
Focus Group, and before that owned by The Attachmate Group, then before
that by Novell.

Both openSUSE and the commercianlly supported SUSE Linux Enterprise are
alive and well.  Obviously the firm doesn't get a lot of press on this
side of the Atlantic, but then, it seldom has since the boom years.