[LHA-misc] You want opinions? (was: Another article ...)

[John <jk...@pr...>]

Yes, I've got some nerve popping up from the depths a launching
torpedoes...

On Wed, 1 Nov 2000, Neil Cherry wrote:

> So where does the project stand (I your opinion, of course)? Don't be
> afraid to hit me hard if you feel it's necessary. I'm not really sure
> I know what I'm doing, I just know what I want!

  There are two words which don't get much play in Linux development
  projects: Requirements and Design.

Wow.  Let's just shred 99% of all SourceForge projects in one sentence.
Think about the successful OpenSource projects in both quantity and
quality.  

Quantitatively, the amount of active projects compared with the total body
of project startups makes the successful ones pretty rare.  Overall
quality is fairly low.  My suspicion is lack of project management.

Without strong management, everybody scratches their particular itch and
rare is the developer who itches for QA, documentation, or effective UI
design.  Weak management never has the will to halt new feature work in
favor of bug resolution when a release is imminent.

Requirements and Design are *excellent* management tools.  Not only do you
know where you are going, you have a fairly good idea of how.  Without
paying these up-front dues you get monolithic, non-scalable, user-hostile
applications.

Proof-of-concept code and RAD tools are good *tools*, but I've seen FAR
too many of these types of projects become The Final Release.  (I'm sorry,
RAD generated code is just not capable of supporting enterprise-level
applications).

  ... avoid writing code too soon!

Amen.  Think about your goals.  Consider that the goals you set now will
be need to sustain the project for the next three years.  The Mozilla
project (rightly) shifted gears one year into implementation.  While it
was the right decision, it cost them 6mo. to 1 year.

My Goals wishlist:

V1.0
- UI agnostic <-- This just saved my butt[1].
- Device agnostic
- Common device properties
- Extensible device properties

V2.0
- Inter-server communications

If these elements are properly defined, scalibility comes for free.  It
provides a lot of flexibility.  As long as somebody can write a driver for
their bright shiny new toy, we can support it.  Want to support telnet,
web, and LCDproc clients?  As long as they can work with the common UI
specification, not a problem.  Yes, yes, it's my 'frameworks' argument.

And to be completely harsh: you want funding?  You need to have a fairly
solid 1.0 release firmly under your belt and be well on the way to
V2.0.  Gone are the heady days of the Internet Gold Rush of '99.  The new
millenium *demands* profitability.  Come, let us build the new Silicon
Valley over the abandoned mine...

Ok, so I haven't had much sleep in the last two weeks...

john

[1] Started out as a Java (Swing) based application.  Performance concerns
have caused us to rethink it as a web application.  Luckily, I made the
front end fairly stupid w.r.t the project.  Sadly, that 33% of my time is
wasted.