Re: Some ideas

[Lally Singh <lally singh gmail com>]

On Wed, 04 Aug 2004 20:42:26 +0200, Angel Costa <angel costa iespana es> wrote:
> 1 - Software engineering:
> 
> I've looked at gnome web site and i've seen the project schedule and how
> does it come true. It's genial that part of the SE is being done, but i
> miss design prototipes for almost everything. I've found a packets
> diagram for the whole gnome project, but not for every part of the
> desktop. For the future (2.10 and beyond) could it be possible to create
> the design before create anything? this will be a very great step to
> improve GNOME. I know it's difficult and for all the programmers it's
> boring, but it's important (i think).

It is, but those skills are rare, and the mindshare for those skills'
importance is even more rare :-(

>  2 - programming language
> 
>  I know C is a very powerfull language, but nowadays, almost every
> software is object-oriented. I think it could be genial to agree one
> "official" programming language, this language must be object oriented.
> I know C# is proposed by a very big group, also Java. My opinion: C#
> because mono is a very ambitious project, but i'm open for the final
> agree.

C's selection isn't so much in technical merit as it is in widespread
use.  While other languages and OO methodology have large, obvious
advantages, one major disadvantage is that OO programming (GOOD OO
programming that is), takes good time to learn and understand.  That
disadvantage cuts down on the number of potential programmers for the
project.  However, one of C's advantages is easy linkage
compatibility; it's easy to link programs written in many other
languages to C.

> 3 - Developping environment
> 
> The best IDE i know for gnome is Anjuta. But it is thousands light-years
> from Kdevelop, DevStudio or even Eclipse. We NEED a powerfull CASE tool,
> not only for write easily our source code, just for help the programmers
> to make the design of the application, look the help docs and write the
> programmers documentation easily.

CASE tools are good, but it'd probably be smarter to leverage what the
open-source folks in Java land are doing instead.  tigris.org has some
good stuff in this area.  Not gnome, but it's open source and works
today :-)

However, adapting Eclipse's CDT into a GNOME development environment
is a great idea.  We could leverage all the CASE strengths in eclipse
and just put in the bits to make it an easy build.  Leapfrog kdevelop!
:-)

> 4 - Write my library and tons of examples
> 
> I've used GNOME libraries for my develops, i can say they are
> documented, but only the api, not the library. I've found tons of
> problems that anyone more had had, but no examples where joined to the
> library distribution. "See the sources" is one of the most phrase we
> have heard, but no comments are found in the code.

It's a problem of motivation.  Only way to really convince people to
change in that regard is by example; perhaps submit a bit of
documentation here & there to the variouss library devs, try and
entice them to fill in some of the blanks? :-)

PHP has a great setup for this, the docs are online, and let you
respond to them as if they were articles on slashdot; that way
corrections & tips are dead-easy to submit.

> 5 - My app is better than yours
> 
> The most power of the open source philosophy is that if i see something
> i dislike, i can change it or even, make it by myself. In this case,
> i've seen some projects that begin with the exactly same goal, each one
> with its own way, but they are give up. for example libgtkhtml, there is
> currently versions 2 and 3, i don't know why. please we must join us to
> improve our products not to make more products.

As long as people are free to make up their own minds, they're often
going to decide against what's best long-term.  Ah well :-(

> (I skipped #6)
> 7 - a little precipice
> 
> I mean with this that it could be possibly to stop everything we are
> developping, and with the ideas i've proposed, begin from 0. Study the
> architecture of the desktop and design it again (enlightenment have done
> it), and, using the new language (C#, Java, C++, Python... but not C)
> implement the whole system: CORBA wrappers,  GTK+, GNOMEUI, GNOMEPrint,
> Bonobo (but now multihost), gnome-panel, gnome-vfs and nautilus (or a
> new gnome-file). (this is my proposal, if you don't like it i'm sorry,
> but we are free to express ourselves).

Honestly, redesigning or reimplementing any system unless absolutely
necessary is almost always the cause of a project's death.  
Redesigning systems from scratch is easy; evolving a system to adapt
to the user's needs over time is the real challenge.

GNOME's got lots of strong points, and some weaknesses.  But, its
competition on other platforms have decades of evolution and
controlled design under their belt.  Open-source projects will
naturally have some sprawl to them.  Different from what people are
used to, but just as powerful, far more resilient, and honestly a lot
more fun than traditional desktops :-)

-- 
Evil will win because good is dumb.