gripes, etc.

[Andrew Sutton <asutton21 home com>]

i spent the greater portion of the day looking for an application that
would let me do something like rational's modeler (class diagrams, use
cases, etc.). this led me on a long quest of installing around 30
different libraries, applications and api's. finally, i gave up, finding
no application that provided the flexibility i was looking for, and
considered writing one for gnome - mainly because there's lots of
re-usable software out there, but found some basic problems with the
gnome conceptual model (for lack of a better description). this is a
brief list of gripes for gnome...

1. there's no roadmap.

i looked all over gnome.org for some kind of a product roadmap. i picked
up the major categories of development (desktop, office and
development), but there's nothing that seems to tie those together. any
developer can walk thru the product list and pick out back-end,
middleware, and user interface software, but it's confusing and time
consuming.

2. i couldn't find any evidence of a steering commitee.

i found a couple of primary developers - but that doesn't seem to be
quite the same. it left me questioning the future direction of the gnome
component model and the directions that various applications are taking.
for example, what's going to happen when corba 3 comes out. it defines a
component model - a very comprehensive one too - i scanned thru the 500
page spec a while ago.

3. no middle ground documentation

every piece of documentation i read on gnome.org and went from screen
shots with brief capability descriptions to header files and
documentation of function calls. the task of learning gtk and gnome is,
for lack of a better word, a little daunting. especially, after coming
home from a long day of coding for different systems. in short, there's
either too little or too much documentation. somebody needs to draw the
object model for gtk and gnome (since they're so obect oriented). uml or
booch would be perfect. the learning curve for new developers can be
significantly reduced (assuming they can figure out the notation).

4. i still didn't solve my problem

just a gripe in general. i think the linux development community could
definitely benefit from a design tool like rational's modeller. imagine,
making the right decisions early can save you from complete rewrites
later (as technologies change and evolve). so, if there's one in the
works... great!, otherwise... um... anybody want to write one? i have
limited (read none) gnome development experience, but i can definitely
contribute solid ideas.

imagine the possibilities... integrated development envirionment with
design, gui development (glade), code generation, documentation
generation, code editing (wouldn't be much of an ide without it),
requirement tracking, bug tracking, configuration management... just
thinking big :)

comments, questions?

Andrew Sutton
asutton21@home.com