|
|
| GeoCommunity Mailing List |
| |
| Mailing List Archives |
| Subject: | Re: GISList: re: Linux RS and GIS software for the masses |
| Date: |
06/10/2002 07:58:37 AM |
| From: |
Dave McIlhagga |
|
|
Seems like I opened a can of worms here. The funny thing is, I'm not
speaking from a philosophical "Open source is wonderful" perspective,
but as a businessman who has tried to find the best value for my
clients. A comparison of alternatives for us in a given scenario has
frequently presented Open Source solutions as the best choice.
Dimitri Rotow wrote:
>
> > applications I have seen. Within the next few months, MapServer will be
> > upgraded to support PDF output, and will be tightly integrated with
> > Flash technology providing the ability to develop fully interactive
> > mapping applications in a flash environment. Obsolete eh?
> >
>
> Well, since you asked, yes. You've actually helped make my point about
> blind spots in the Linux community. MapServer is obsolete for two reasons
> in the context of this thread ("GIS software for the masses"):
>
> 1) It cannot be configured without substantial programming. In that respect
> it is a typical Linux appliance bereft of any awareness of how important
> interactive interfaces are for mass usage. It is a typically Linux blind
> spot response to rise to the defense of MapServer without addressing the key
> point, that is cannot be set up without substantial programming.
True in the past, but wrong in the very near future. I too have seen
this deficiency in the stereotype thinking from the Linux world.
Because of this, we have recently created an add-on to MapServer that
allows for publication of mapping applications entirely through a web
browser through a nice user-friendly interface. It is currently in
beta-testing mode, and when released - as open source - will make
MapServer very useable by the non-techs of the world.
> 2) MapServer has zero interactive GIS capabilities, so you have to purchase
> a real GIS to get your data squared away. The idea that one should have two
> GIS packages, one to edit and arrange one's GIS data interactively and
> another one, a perverse server engine with which one must communicate by CGI
> scripts, to publish that GIS data to the web is a very obsolete idea.
> Leaving WYSIWYG issues aside, it's a dreadfully obsolete idea to think that
> one should have to learn two radically different GIS approaches (with all
> the attendent risks of getting things wrong in the handoff) just to publish
> something to the web.
>
> The internal technical architecture of MapServer is also obsolete compared
> to what one would do with a serious programming effort using modern
> approaches. However, as my comments about the abacus in my earlier posting
> indicate, just because it uses old methods doesn't mean it can't be used to
> do cool tricks. If you apply enough time and money to it and are willing to
> burn enough computer cycles, sure you can create fine web sites with
> MapServer.
I think you're forgetting some very important issues here. First of
all, contrary to what the traditional GIS industry thinks, the real
benefit of Internet mapping technology is not to put GIS on the web, but
to seamlessly include a spatial component into Internet applications.
That requires very different technology then a Desktop GIS. Why bloat
an Internet engine with a whole pile of functionality that is
appropriate in GIS that 99% of web applications don't need? Just as
Desktop applications are designed and built to do appropriate tasks on
the desktop, so is an Internet engine. in this case, the main focus of
MapServer is to provide a ton of capabilities for designing high quality
maps, while accessing data from a variety of sources - not to do
buffering, data conversion, vector editing et... which are appropriate
on the desktop.
Also, the technical architecture of MapServer is designed for efficiency
and robustness, two very important issues for a server application.
Using a "Modern approach" does not inherently make it better in all
circumstances. MapServer from our perspective has been very bug-free,
fast, flexible, and supported very rapid deployment of applications. +
being open source, it has allowed us to do quick fixes or add-ons as
required. It's been like a piece of commercial software with added
bonuses. What more could we want? Short on documentation - yes, short
on user-friendly tools - yes, but I think what I mentioned above will be
addressing this.
> I strongly agree that open source projects are not dead end projects. The
> key question is why open source projects, in Linux in particular, have not
> resulted in any world-beating interactive applications. I gave my opinion
> on that in the previous posting, but I'd like to hear your opinion... why is
> it that Linux and open source efforts have not produced any significant
> interactive applications that have achieved anywhere nea
|
|
Sponsored by:
For information
regarding
advertising rates
Click Here!
|
