[Sussex] Sky boxen / Embeded LINUX and NT

[Steve Dobson]

Geoff

On 27 September 2002 at 12:17 Geoff Teale said:
> I thought this one might bring forward some interesting points :)

Didn't want to let you (or my public lol) down.

> The question, as you recall:
> 
> > > How can we be sure that LINUX in embedded space isn't a lame 
> > > duck just like Windows?
> 
> Steve answered:

<snip comment="where we agree">
  
> > 2/. Linux is driver by technical goals not by commercial needs.
> > Look at the time between major releases of Linux to WindowOS.
> > Does anyone not see that announcing that the next OS (in '93?)
> > that the next OS will be called Windows95 doesn't put a 
> > commercial pressure on the OS developers to release in '95
> > regardless of it bug status.
> 
> Hmm, this is a great argument, but the reality is that a lot 
> of embedded development is using kernels that are heavily
> 'tweaked' by companies like (i.e. Lineo Embedix) who _are_ a
> commercial venture and _do_ work to release deadlines. 

I take issue with the "heavily tweaked" argument.  The GPL doesn't
allow anyone to modify the source without publishing the source
to those changes.  And that fact that those companies are 
advertising that it's Linux strongly suggest to me that they are
playing by the GPL rules.

If there is a company making such tweaks then the RT/Embedded
OS community will be tracking it and rolling the "good" stuff
back in to their public RT patches to the Linux core.  These are
just as driven by technical goals as the Linux core; but
admittedly different goals.

I would also suggest that most of those "tweaks" are just the RT
patches and some of those have been rolled back into Linus's tree.
[Note: that I say Linus's tree - there is no one source for Linux.
Alan's releases are just as valid as Linus's.  It just seams that
most people take from Linus's releases.]

I am not aware of any RT/Embedded versions of Linux that have
features that ONLY they offer (ignore device support I'll come to
that next).  Therefore the "value add" of the companies is doing the
donkey work of configuring and building the OS for a given platform.
As they are not adding features (that aren't public domain themselves)
THEY are not "tweaking" the kernel, and therefore my "technical
driven" argument stands.

"Technically driven" still standing.

Device support is about the only place where rights to "new features"
can be retained by a company.  But these have to be done as loadable
modules under the GPL.  If they are compiled into to a single image
then they become "infected" by the GPL.

I don't count driver modules as "heavy tweaking".  This is because the
core/driver APIs are strongly defined and therefore it is possible to
"black box" test a loadable module.

Also, and probably more important, loadable modules don't add features
to the kernel.  I mean features like scheduling, VM management, caching,
etc...  When selecting an OS (for RT, embedded, desktop or server)
hardware support is important.  But it is just a tick in a box.  Once
you have eliminated those OSs that don't support the hardware you have
the choice is then down to features and cost.

On of the reasons of selecting Linux or WinCE over more traditional
RT/embedded OSs [VxWorks, LynxOS,...] is that the development and
target environment are (basically) the same.  Given what you said
in your reply to my reply to your e-mail Linux wins here as WinCE
is a derivative of Win95 nor 2K or XP.

"Technically driven" dented [just], but still standing.  I look forward to 
your reply.

<snip comment="where we agree">
  But thanks for backing up my argument with good examples.
</snip>

Steve