No subject

[Stephen Lime]

Hi:

> Hello,

>  I have been working on a system similar to 
> mapserver for about a year, and was pleasantly > supprised when I came across the mapserver 
> website.  I have a few issues with the server 
> in it's current state, and am wondering if this > is an open project or mostly a personal one 
> which happens to be open source.

In the past I would describe the MapServer as more of a personal project. However, I'm very interested in seeing it go full blown OpenSource. That is the plan anyway after the next version is released sometime in October. We are working now to enable a formal CVS with web access in hopes of creating a formal (and distributed) development team.

>  One of my main issues with the system is the 
> degree to which it depends on the gd library 
> (and thus, the indexed image format used 
> therein).  In my own shapefile renderer, I have > been using a 32bit RGBA array as my image 
> format, with functions to load/save from 
> various graphics/spatial data formats.  In 
> reguard to the recent discussion about 
> distributed rendering of image planes, with the
> master server slapping the planes together, 
> this could be accomplished much easier if the 
> individual distributed renderers did not have 
> to worry about using up a colormap.

I agree. Version 3.2 has hidden all the colormap manipulation from the user and requests color information as RGB values. This is in preperation for a 24 or 32 bit image format. I've looked for alternatives to GD and haven't found much. Actually ImageMagick has GD-like functions but the price is steep as the author reports 4x speed decrease. Perhaps this is an area you could contribute, sound like you must have solved the problem. Again, speed is critical.

> Several other design issues, as well as 
> questions of extensibility (new data formats), > have led me to question whether a complete 
> redesign of your system is in order.  I have
> been trying to decide whether I wanted to
> attempt to merge the mapserver layout/parsing
> engine with my own code, but I wanted to get
> the opinions of those actually working on the
> mapserver code first.

Complete redesign? Not without some real good reasons, but I'm listening. The rendering engine has been redesigned in version 3.3 so that integrating handlers for other data sources is *much* easier than before. Supporting every format under the sun is real low on my priority list. It's not like conversions to shapefiles are hard any more so I've no problem with using that format. I'm more interested in generic access to distributed spatial data. OGDI (www.ogdi.org) really peaks my interest here. Non-standard spatial data access via a scripting environment like perl is also high on my list. 3.3 will introduce MapScript a perl module for creating maps using the MapServer API.

>  Are there any current plans to redesign the
> mapserver system?  If so, will it be open for
> discussion among those using the system?

In an OpenSource environment I certainly hope that
users will want some input. With 3.3 I hope to have a nice platform to build on. I'm certainly open to discussions on how we might make it a better product.

> I will go through the mailing list archives to
> see if any of these questions have already been
> brought forward and answered, but would
> appreciate any response I can get.

Most folks are just getting started and aren't that concerned (yet) with questions about the internals. You may find some interesting reading though.

> Thanks,

> Aaron Stafford
> ender at gic.csuchico.edu

> Geographic Information Center
> California State University, Chico

Steve