Christian Lavoie wrote:
>
> Actually, I think more and more people are wanting Microsof-like
> applications, because the Microsoft philosophy has some good ideas,
> especially when you are a end-user.
[...]
> Let's take the Windows' IE and Office integration as an example. In the
> basic, it's a great idea. You get to do everyday tasks more easily, the
> appropriate tools are more handy and more and more softwares can use those
> apps as subsets of themselves. You get a serie of to-be-powerful tools that
> are imposed as standards, guaranteeing that your knowledge is to be
> preserved from task to task.
>
> Now my point is: Microsoft has some great ideas, and it would be a shame to
Well, I disagree. Personally I dislike massively integrated
applications like Outlook and Explorer. They are too big, too slow
and too complicated (to use and maintain). And I'm convinced that my
knowledge of how to use Outlook will be obsolete in a few years
because MS (or any molopoly) has a vested interest in *not* having
standards or letting thier tools be integrated by someone else.
I'd wager that the single biggest reasons that Unix has survived and
prospered over the last 20+ years is not because of it's design as an
OS but because of the software design philosophy of it's interface.
Paraphrased, it goes something like this:
Create small programs that do a single task and do it well.
Support a common communication mechanism so that each of these
small programs can be used together to solve complex tasks.
Take 'ls' as an example. It does not have a -p flag for stopping it's
output every screen - because that is not the task of a "list"
program. Pipe it's output to your favorite pager which will have all
the bells and whistles which a paging program can specialize in.
One might argue that this works for command line interfaces but not
for GUIs. I would still disagree. The communication mechanism will
certainly need to be more complicated than Unix style pipes reading
and writing ASCII but the payoff of having a collection of
specialized, relatively easy to maintain, tools will be worth the
effort. Although I never used an Amiga, I've heard that it had a
scripting language for exactly this purpose. You also don't need to
know ahead of time what your application will be used for. Build the
walls, doors and windows separately and you can build (and change) any
building in may different unpredictable ways.
This is certainly the future direction of software development, once
"integration" and obsfucated standards as a business model (which a
monopoly will certainly want to employ) are overthrown, or more
likely, crumble under it's own weight and complexity.
Just *my* 2 cents...
Keith
