(Apologies for the delay.)

>If the amateur masses with a small number of ineffecitvely
>managed full time engineers can produce a product that
>has been able to threaten a well engineered product such
>as Solaris, how would you propose responding to that?

Hmm - what innovation? Actually delivering stuff 'free' on
such a scale is innovative, but pretty much all of what's
in the Linux kernel at the functional level has been
delivered by Solaris, AIX, and major ISVs before.  Or
BSD.  Or Windows.

There is continuous re-engineering, but its not clear
to me that this is innovation, rather than reaction to
scalability barriers inherent in earlier 'design decisions'.

How would I respond? By managing resources effectively?

Look at the duplication and the number of herculean
efforts that never get merged.  Its shameful. I wouldn't
consider contributing to that for pleasure.  Good thing
most people contributing to the kernel (in non-trivial
ways like USB and PCI signatures) are paid for it.

My concern in making Solaris development 'linux-like'
will give up a valuable management role.

Shuttleworth has illustrated how much value there is in
strong mangement and leadership, and my concern is
that by trying to join the 'developer lead' ranks of
'just show me the code' cowboys Solaris will lose.

If you must, then at least:
 - 'show me the customer requirement'
 - 'show me the technical design'
 - 'show me the code'
 - 'show me the tests and instrumentation'

(Oooh look, the c-word.  And not 'code')

Strong management and leadership is important.  And
that needs empowerment, authority, and an ability
to wield resources.  It doesn't imply that volunteer
or external resources aren't used, but they have to
be reliable.

>Especially if a large portion of the market (ie the level
>req'd to support bringing Solaris to you) decides that
>near enough is good enough?

Is that really what you want Solaris to be?  Near enough?
Don't give up on Solaris' strengths.

>If the amateur masses can drive innovation and pursue
>bringing more features to OpenSolaris, isn't that something
>to be pursued and capitalised on, especially if it acts
>as a gap filler?

Gap filler?  Is that it - just a catchup game and then
something else?

I think its clearly necessary to distinguish what the
development processes are for:
 - actual kernel and libc core
 - packaging into specialised distributions
 - userland stuff

Linux development in the first of these has tended
towards professional now, by companies with vested
interests.  Some of the companies doing this are
Sun's competitors and won't contribute to Solaris on
any basis because of brand association.  So don't
expect real help here except from people doing
specialist consultancy who can perhaps help with things
like the bootstrap on Intel Mac - most of the heavy
lifting will have to be done IN Sun, or perhaps
sponsored by Sun.  Don't kid yourself that any amount
of openness is going to change that.

The packaging is largely an issue of:
 - build hosting
 - documentation

(Its not clear, for example, how to make a PXE boot
client that then uses iSCSI to access swap and private
workspace: Sun *has* a diskless boot system but its not
what I need. It should be fixable.  I can make Ubuntu
do that - but the swap will deadlock and what you get
is Linux engineering at the end of the day)

The userland stuff should come from userland emulation
of Linux over and above POSIX to make the masses of
badly engineered stuff work more easily.  If you could
get Solaris to install easier on ordinary Joe's hardware
and give it away on news-stands, its a problem that will
(slowly, perhaps) go away anyway.


And university researchers?

They do their own thing.  By and large, research has
a goal of proof of concept only.  Its NOT 'engineering'.
Its 'R', not 'D'.  Don't overestimate the value of 'R'
if you can't follow up with real 'D'.  The 'R' _is_
important - but not to a product.


>>So? Unless they manage a datacentre or otherwise buy
>>or specify, they're just Joe Public.  Do you really *need*
>>the ones who have that responsibility and flunk it to choose
>>based on years-ago amateur hacking relationships?
>>
>
>Wow.  You do realise that this is an incredibly over
>generalised statement and that it insults a lot of people
>you've likely never met, never mind understood what
>they've done or do?

What's the Solaris user base?  Who buys Sun kit?
How do you expect that to change?  You can't afford to
lose your existing base.  Make the omelete, break the eggs.

>I don't know any 'freedom' zealots inside Sun but I'm
>sure they're out there somewhere.  The point of that
>is to say that just because someone is an open source
>developer doesn't mean they are a 'freedom' zealot.

I didn't say that.  But Linux development is full of it,
and you seem to want to ape that.  Which throws away a
valuable differentiator.

>A lot of people I know would be quite happy with binary
>blobs, so long as they could be used on their platform
>of choice.

Yes, practical people.  Make sure that Solaris is very
directly aligned with their needs.

>It sounds like your expectations of what an open source
>product should be and what they actually are aren't in
>100% alignment.  Rather than say one or the other is at
>fault, I would recommend that you continue using Solaris
>or other products that meet your needs.

The point is - as a consumer I DON'T care what the development
model is.  You are in the trap - you see 'its an open source
product' and to you that's part of the product and should shape
the user expectation.  But to us out here, we see a DVD image
with stuff to install.

Be careful - do you really mean to imply that my expectation
of an open source product should be lower than it would be
otherwise?  Is that what Open Solaris is about?

James

-- 
No virus found in this outgoing message.
Checked by AVG Free Edition.
Version: 7.5.446 / Virus Database: 268.18.15/728 - Release Date: 20/03/2007 
08:07
 

_______________________________________________
opensolaris-discuss mailing list
[email protected]

Reply via email to