[cgl_discussion] Linux users and the Linux community

John Cherry cherry at osdl.org
Wed Aug 4 08:01:49 PDT 2004

Gerrit recently posted to the data center linux group regarding 
the kernel community and bridging the gap between linux users and
linux developers.  Gerrit references Andrew's OLS speech, which is
also excellent in describing how to bridge the gap between linux
users and linux developers.  It is a good read.  You may not agree 
with all of it, but it will make you think about the issues.



I think it is well worth noting that communities are built of
individuals and often of multiple sub-communities.  The same is true
for the linux kernel community.  And, while there was a lot of negative
commentary at kernel summit and on the mailing list prior to KS, the
"good" news is that this means that neither of these communities is trying
to ignore the other.  That at least provides the basis for ongoing

Many non-kernel communities are simply ignored if they don't cause
any pain and they don't make a compelling case for inclusion.  One of
OSDL's main problems is that they have been effective in inflicting
pain on part of the community while getting pleasure from the vendor
marketing/sales and OSDL affiliate aspects.  Specifically, CGL has
been a point of repeated discussion.  Red Hat, and, to some extent,
SuSE have been thumped repeatedly with the CGL specs by their customers.
Both companies are relatively small compared to most of the vendors and
customers involved in CGL and have never seen the level of customer
cooperation that a standard or specification generally achieves.
The customers are happy with CGL because they can point to it and say
"see, all those things you were arguing with us about that we couldn't
state clearly enough for you to understand - here they are".  And,
RH/SuSE then have to walk through an formalized list and point out how
many could be done differently.

Well, if that was lever that was needed to help bring the community
to the discussion table, so be it.  It isn't pleasant, but most of the
organizations involved now are companies with customers.  Making those
customer happy is what most of the vendors want to do, because happy
customers spend money.  ;)  The problem, of course, is now some of the
community members come from RH/SuSE, etc.  And, they share their
experiences with the rest of the community.  So, we start with a negative
bias from one of the key players, which we need to make every effort
to address in future projects, e.g. DCL and DTL.  And some of those
steps have been started, e.g. opening up the SIG lists and encouraging
appropriate community members to be active.

The second significant problem that OSDL has with the community as I see
it is that OSDL has made the claim that it is trying to be the center
of gravity for Linux.  Many of the core community enjoy the lack of
centralization and control of Linux and see OSDL as threat to a Linux
"of the hacker, by the hacker, for the hacker."  The problem is that
nobody wants RH, or SuSE or Novell or IBM or HP or... to own Linux,
but in the real world of customers, customers want to have *someone* to
talk to.  The kernel community has not really understood this aspect,
and as developers, they generally have no desire to understand it.  Of
course, this is often because the Linux community as a whole has very
little understanding of the people that really use Linux today.  There
are always individuals with specific experiences inside the community that
know of a customer or a user of Linux.  Most tend to extrapolate from their
own experiences that all users look like "This" (pointing to the few that
they know).  The reality here is that there is a culture clash between
the vendors experiences here about what customers look like and what
the development community perceives linux users to be.

The general answer to this problem is to spend more time bridging the
gap between linux users and linux developers.  I think (I hope!) that
having Goldman Sachs and, to a lesser extent, Amazon at kernel summit
helped provide a little bit of the vendor view of what a customer looks
like.  And, I think Andrew Morton's keynote provided a clear message
to the vendors (and even the vendor's customers, who mostly weren't
listening, unfortunately) about how vendors should work with the

I'm hoping that involving customers, ISVs, and developers (kernel and
otherwise) in the appropriate SIGs will help bridge some of this
difference in cultural background.  I think some of the other suggestions
made here, e.g. educating customers better on how to work with the
community, is another place where OSDL can add value.  And, opening up
the "capability" documents for earlier and earlier review may also
help over time so that people can help develop the context over time and
have their input more directly into the process.  The vendors also need
to understand that "that's the way we've always done it" is a very bad
response to the linux community, and often doesn't serve the customer very
well either.  But at the same time, when a major customer says "I depend
on Java", the Linux community response can't be "Well, tough, Java stinks".
Some things are business or market realities that aren't going to change
over night.

I'm going to strongly recommend that if you've read this far through my 
drivel, you go read this and internalize this speech from Andrew


It is a roadmap for at least the first few steps of a better partnership
between vendors and the kernel community.  And, it may contain some very
sage advice for how OSDL groups work with the community as well.


More information about the cgl_discussion mailing list