Re: Where we stand in regard to the future platform / desktop technology

[Sean Middleditch <elanthis awesomeplay com>]

On Mon, 2004-03-29 at 03:25, Rui Miguel Seabra wrote:
> On Sun, 2004-03-28 at 17:55 -0700, Ryan McDougall wrote:

> > Thats why we restrict ourselves to ECMA, and our *own* library stack. No
> > one wants GNOME to depend on .NET -- its simply not on the table. Please
> > read the big long parent thread if you don't understand what people are
> > arguing.
> 
> The question is that it is incomplete enough to be useless to a lot of
> people, so not really that usefull to be something to base upon.

> > Thats the whole crux of the discussion! Please let us know what you
> > consider the alternatives to be.
> 
> perl, python if you want to make a smaller selection.
> python might be more apt since it has a simpler syntax and is object
> oriented by nature, which adapts well to the GUI paradigm.

You're missing one of the main arguments behind using C#/Java - that's
what's being taught in schools, that's what companies are looking for in
the coders they hire, and those are the languages people are *expecting*
to use.

Python is great; I use it and love using it.  But then, I'm a hard-core
programmer, not a corporate monkey (no offense Ximian folks) with no
experience but some college Java classes who expects to slap together an
app and push it out the door.  Which, no matter how much we may dislike
it, is what the real-world of corporate programmers is generally like.

And don't also forget how very unstable Python is in regards to
modules.  A user upgrades Python (say, upgrades their OS?) and *poof*
half of everything stops working properly.  You need a copy of every
module installed for every version of Python you have (because your apps
are generally made in a way that invokes a specific version of Python,
again to deal with broken module versioning).

When Python commits to a stable release process that provides backwards
compatibility for *all* modules, scripts/programs, and usage, then it
would be viable.  Until then, it's only a pain for users.  Especially as
soon as you require anything more than the core Python libraries (try
finding the right RPM/whatever for the right OS for the right Python
version and all its relevant dependencies) and then only those which
were available in 2.0 (so you're close to backwards compatible), which
effectively reduces you to about as much functionality as depending on
ECMA CLR plus the GNOME libraries.

Microsoft worked very hard with the CLR to avoid those versioned
dependency problems that Python suffers from, and its a big selling
point for the CLR.

The only distro I've seen that even *attempts* to handle the Python
versioning problem is Debian, and that still won't solve anything for
third-party or application-specific modules.

> Rui
-- 
Sean Middleditch <elanthis awesomeplay com>
AwesomePlay Productions, Inc.