Subject: Re: business case for mechanized documentation
From: Rich Morin <rdm@cfcl.com>
Date: Fri, 7 Apr 2006 10:01:40 -0800

At 12:41 AM +0900 4/8/06, Stephen J. Turnbull wrote:
> (1) To what extent can OSS hackers have the impact on (say)
>     Solaris that they can have with Linux or a *BSD?

Any time a vendor provides support for an OSS project, there are
going to be effects on the "balance of power" in the community.
I recall RMS expressing concern (way back when) about whether
Cygnus Support would develop GCC (etc) in the wrong directions,
for the wrong reasons, to please their paying customers.

In cases such as Solaris, this gets carried to an extreme.  The
code base is huge and development is totally dominated by Sun.
Even if they work diligently at opening up development, Sun is
going to "own" the code base for the foreseeable future.

However, I think that this misses the point.  Sun has (I presume)
motivations for going Open Source with Solaris, Sparc, etc.  I'm
not sure what they all are, but allowing external developers to
understand (parts of) the code base surely must be part of the plan.
Easing this involvement through better internals documentation seems
like an appropriate move, regardless of any other considerations.


> (2) How big is OOo, something that I'd think one talented hacker
> could have visible impact on?

One of the interesting things about Open Source is that a talented
hacker can make a small, innovative change that has substantial
effects.  For example, the Semantic MediaWiki effort involves very
little code (relatively speaking), but its adoption could transform
the behavior of MediaWiki (and thus Wikipedia).


> 10 programmers vs 5000 staff (maybe not the same units!) on
> Solaris sounds like a drop in the bucket, but if you consider 20
> staff (including managers and support) for docs vs 100 staff for
> the whole project, that's a big hunk.

Some of the benefits from mechanized documentation accrue without
any specific effort on the part of developers.  If Sun had tools
that automagically generated and published detailed documentation
for all of the C and Java source code they ship, even tiny projects
would benefit.

However, proper (IMHO) use of mechanized documentation tools needs
some effort on the part of developers.  Clearly, this uses developer
time; whether this is justified is a management decision, based on
resources and perceived benefits.


> (3) Can Sun et al let control slip away, or are they going to
> need to dedicate further staff to the care and feeding of hordes
> of volunteer hackers attracted by the documentation?

As noted above, I don't think any external party is likely to "fork"
Solaris any time soon.  Or were you driving at something else here?

If Sun wants to reap the benefits of Open Source, they have to make
it easy for external developers to get involved.  Documentation, as
you note, is only part of this.

However, bear in mind that Sun may be more interested in allowing
external developers to "scratch their own itch" than in harvesting
the resulting changes.  If an external developer can solve a problem
without Sun's involvement, Sun saves on support costs.

<historical aside>
Many years ago, DEC supplied the BLISS source code for VMS to some
of its customers.  However, the distribution was on microfiche and
no tools were (IIRC) available for recompiling and rebuilding the
OS.  So, the external developers could only do binary patches (!)
on their installed software.  Even so, many sites bought the fiche.
I wonder how many patches ever resulted from them...
</historical_aside>


In any case, let's not get lost in a discussion of Sun and Solaris.
Novel and Red Hat have similar support issues for their versions of
Linux, as do other large projects (eg, Apache, Mozilla) for their
non-OS offerings.

-r
-- 
http://www.cfcl.com/rdm            Rich Morin
http://www.cfcl.com/rdm/resume     rdm@cfcl.com
http://www.cfcl.com/rdm/weblog     +1 650-873-7841

Technical editing and writing, programming, and web development