Sorry, I'm not really going to be able to respond to the whole thing,
but I'll start with some of it.

I've been around a few years watching NetBSD grow since about 1.4.2.

But I speak for myself. I probably will make a few mistakes in what I
say and I hope people will correct me. Warning: What I say isn't
necessarily how it is, it's just how I see it.

On Wed, 09 Feb 2005 13:19:22 +1300, Joel Macklow
<joelm@webservices.net.nz> wrote:
> And here are my thoughts (for what they may be worth).  I felt compelled
> to write this after reading Christos comments in the annual report.

So have I, but I don't think I have the eloquence to even be in the same room.
 
> 1. "Of course it runs NetBSD"
> Yup, true. NetBSD runs on everything, but as alluded to by Allen Briggs
> in the annual report, some of these have been abandoned.  Should they
> really be promoted?  As I see it, the people we want to attract to using
> NetBSD are data centre admins and CIO's.  NetBSD's biggest chance is to
> have it installed on new machines going into an enterprise.  There is
> also a big opportunity in migration path too.  A marketing pitch is
> being able to say to a data centre: start with i386, then move to
> Opteron AMD64 then POWER5/Sparc 64/Alpha, or whatever.  Multiple
> platforms, One OS. Then they can leverage the knowledge base they have
> developed as their data processing needs increase.  This is not an
> altogether stupid idea, Sun are doing this exact thing with Solaris 10.

There's nothing much lost by losing a platform. The port maintainers
are the ones who work on it, not the core team (well, that's not 100%
true, but it's good enough for now). If they stop developing for that
platform, then either someone else picks it up or they don't. The
small part of what is lost is keeping the MI code honest with yet
another wacky platform, or something. And no, it isn't a stupid idea,
the world is changing but NetBSD stays mostly the same. This is a
virtue.

> But here is the challenge for a new NetBSD admin.  I have this
> (hypothetical) new P4 box with an 925X and ICH6R chipset and a bunch of
> SATA drives.  Will NetBSD install on it and work properly?  I can't find
> that out from the web site.  I have to trawl through the mailing lists
> to find out if anyone else has tried.  This is not user friendly in the
> slightest.

What you are talking about is essentially hardware verification
testing. This requires $$ and doesn't buy NetBSD itself anything in my
opinion. This is where a 3rd party vendor can contribute by spending
the money to do this for a customer. And it would be really nice if
they decided to divulge this information. If they need help getting it
working, they probably will. This is exactly the way it works with
Linux, in my experience. You'll see some of the "large" linux vendors
doing some of this, but most of it is done by the implementation
people or hardware vendors, whoever they are.

> What about the new POWER5 machines from IBM.  Will it install / work on
> them?  I don't know the answer, and I can't seem to find out either.

Probably not, for the same reason the G5 Mac doesn't work yet (it
still doesn't, right?). This might be useful to post somewhere, but
I'm not sure that NetBSD is into saying what isn't supported.

> Alpha?  I get the impression that it works real well on entry level
> gear, but doesn't fully support all the high end boxes. Clear knowledge
> regarding these things is required in order to be able to recommend or
> decide upon Multiple Platforms, One OS.

But why? Where do you get this with Linux? NetBSD doesn't have some
large profitable company behind it to be able to do these things. This
is where enthusiasts that own these boxes help out. This is a
volunteer effort. Do you have some of these boxes? Can you kindly try
to install NetBSD and tell us your findings? I have a few boxes at
work that I've installed on, and I decided that NetBSD "just works" on
them and I had nothing to add.
 
> Possible solution:  Web site pages that detail the LATEST hardware
> platforms and chipsets supported.

I think that most try to do this, but again, it's a volunteer effort.
Some of this info is very out of date, I have to agree.

> 2. Reference architectures / usage guides.
> MS does this:
> http://www.microsoft.com/windowsserversystem/solutions/default.mspx
> Sun does this: http://www.sun.com/service/refarch/
> NetBSD has?
> 
> This kind of information is ideal for IT departments looking to solve a
> particular problem.  A guide matching hardware and software together for
> particular tasks.  Maybe a box to do AV/SPAM scanning before hand-off to
> an MS Exchange box for example.
> 
> Possible solution: web site pages that contain reference architectures
> and pkgsrc packages to solve each problem.  Maybe even some new
> meta-pkgs that will install and preconfigure a solution to a problem.

You're running a standards compliant Unix box. Solutions can be found.
Someone needs to test them. This requires $$ and time. Again, who is
going to do it? You're a Linux guy, I'm sure you've done this before?
You just gotta be savvy and make it work.

Maybe we could have some kind of volunteer effort for people to write
in about their configurations.

I sure could. I have a few boxes doing various things, but they are
rather pedestrian in comparison to what I think you are talking about.

> 3. Datasheets
> Thunderbird has this:
> http://www.rebron.org/mozilla/stuff/datasheet-thunderbird.pdf
> 
> Earlier this month, someone said that they were going to an event and
> wanted some datasheets to hand out.  They were referred to the advocacy
> page on the site and suggested that they update some of the sheets.  I
> would be happy to have a go at generating something like the Thunderbird
> one, but what do we put on it?  This refers back to point 1.
> 
> Yes we can talk about the clean code base, pkgsrc and the portability,
> but a potential new admin wants to know specifics (benchmark graphs for e.g)

Again, see my point about large profitable organizations, etc.

> 4. Web Site
> Now I know I am being a glutton for punishment here but....
> 
> I understand that the underlying structure is in the middle to migrating
> to DocBook XML.  And props go out to the www team for all the work they
> do! The netbsd.org website has been constructed by tech support type
> people not marketers. The vast majority of the information on the site
> is to do with NetBSD itself....not advocating/selling people on why they
> should use it.  I am not a marketer, but surely in the global NetBSD
> community we must know some friend of ours that is one and earnestly
> desires to do some valuable community service!  If the dev team were to
> tell the marketers what the capabilities of NetBSD were, I'm sure they
> could easily come up with something very impressive in terms of
> direction and wording.
> 
> The way I figure it, MS have spent millions on their site, designers,
> focus groups, user evaluations....  So have Sun/HP/Dell/Apple et al.
> What can we learn from evaluating their sites?  Should we set up a group
> of 5-10 people to perform evaluations on these or other sites and see
> what they come up with?  I am happy to be involved here.
> 
> Then a new web site look using css could be designed using the new logo
> utilizing the information gleaned from the analysis of the other sites
> and from the marketing people we round up.  Perhaps the current site
> could become a technical reference site.  There is no need for a
> new/prospective user to
> search mailing lists for example, but this is very valuable when sorting
> out a thorny problem.

Here's the way I see NetBSD proliferating. Your a xx(x)ix admin and
gee, doesn't this system suck? Wouldn't it be nice if it did X like my
box at home running NetBSD? Maybe I could convince my boss to let me
turn junk box X doing easy job Y into a NetBSD box? This is where I
am, but I seriously doubt I'll be able to work in NetBSD at work, even
though everybody knows that I like it (ad naseum).

The embedded community seems to be using NetBSD and it seems to be
proliferating, however rather quietly. But I think it is. It's just
hard to tell. This is good.

The big bold server people seem to need big bold names to go along
with their big bold servers, and NetBSD isn't there, yet. But it's
coming. Things like this internet land speed record don't hurt at all.
But this stuff you are talking about above with solutions,
verification, etc are what generate sales. NetBSD can't do this. Not
yet anyway.

You seem motivated. This is good. I wish I had more time, I'd probably
be volunteering for the same things you are. But I'm also still just a
student of NetBSD, trying to learn and help out where I can. Not sure
I'm ready yet.

Andy