[lsb-discuss] LSB conf call notes for 2008-07-30

Alan Cox alan at lxorguk.ukuu.org.uk
Tue Aug 12 00:59:59 PDT 2008


> terms of practical problems, and not just religious issues.  The

What is this a worked example from Machiavelli ;) Belittle the alternative
view point 'religious issues'

> issues you have and the issues that Mr. Herrold have are quite
> different, as near as I can tell.

'divide and conquer'

> As far as trademarks are concerned, the Linux Standard Base(tm) is a
> trademark, and we operate the LSB under very similar rules as Java

Pick an irrelevant apparent counter example...

> does; you can only claim LSB compliance if you pass the certification
> process.  

The examples you give are bogus. Firstly the LSB trademark determines if
you can call your system "LSB compliant" not if you can make it so.
Secondly the other examples you give are products not standards.

> As far as the test suite not being under the GPL --- I'm aware of the
> issue of what happens if Sun goes out of business, and/or a hostile
> takeover of Sun takes place by a bad actor such as Microsoft; as I've
> said earlier, that's something we should work with Sun to overcome.  I
> also understand that some people have a religious objection to any
> software which is not released under the GPL; but I'm trying to

You don't need to work it with Sun. You just need to stop entangling the
basically open LSB standard with the highly controlled (and I know Sun
have good reason for that) Java one. The basis on which you are working
is flawed.

> Finally, I hear your proposed solution, which seems to be some kind of
> to have an optional component of the LSB, possibly with an separate
> certification mark if that optional component is included.  So your
> suggestion, as best as I can understand it, is to have both an "LSB"
> and an "LSB+Java" certification marks and certification programs.
> There are costs to having two marks and programs, of course; the cost
> of confusion over the two marks, for distributors, ISV's and end user.
> The question is what is the benefit of having two marks; or put
> another way, what are costs of *not* having two marks and only having
> Java being included in the LSB.  So far, the practical objections seem
> to be vague enough or not substantiated well enough to make the
> cost/benefit tradeoff argument clear.

There I disagree.

There are distinct marketing advantages in two marks and in co-branding
products and your claims don't appear to match what I've heard from
marketing people.

There is a distinct *advantage* in having the two independant marks and
possibly a co-branding phrase

- Customers see two 'standards' brands. Two is better than one at a
certain mental level
- Customers looking for a java platform see a Java branding
- Customers looking for Linux see a Linux branding
- Customers looking for both see a co-branding that instills confidence
that someone will stand up and not play "its his fault", "no their OS is
buggy" support tag with them.

If LSB 5.mumble happens to include java an LSB branding will be
meaningless to the large Java community. Now when you remember that a lot
of selection decisions are effectively made by non-technical people that
happens to be really important. They won't go "oh LSB 5.x, yeah thats
Java" they will go "must have java logo on it"

So in practical terms as a vendor you are going to need to have two
certification marks and brand logos on the box anyway to sell the
products.

A useful side effect of this is in the open standards plays for
government where an LSB as it standards probably counts as an open
standard, but java as a vendor controlled standard does not. In those
cases being able to say "LSB" would still be valid IFF it doesn't force
in and inherit the java spec (as one government standards proposal puts
it elegantly 'standards must be recursively open').

The two certification programs is also mostly smoke and armwaving. The
costs actually go down for the LSB because the Java standard
administration remains entirely devolved. It goes down for LSB
implementors because they can skip the Java stuff.

The sole requirement on the LSB side is "has /usr/bin/java and can post
us a copy of their JCX test pass statement". That isn't an expensive
business.

> Hopefully folks will find this a fair summary of the discussion to
> date.

Not really, no. But then presenting your position as the middle ground is
also from the same book as we started from.

Alan



More information about the lsb-discuss mailing list