CGI vs. WMS

[Jim Cser]

What are the pro and cons of using WMS over CGI MapServer?   Does anyone
here have any direct experience with both?

We have a Cold Fusion web application that displays a few maps, with
only simple pan/zoom functionality needed * it's not a "GIS-like"
site.  Our previous MapServer apps have used CGI MapServer, but for this
app we decided to experiment with implementing a WMS service.  We're a
CF shop, otherwise we might try PHP or Java mapscript.  Our map layers
are coming from PostGreSQL/ PostGIS, for what it's worth.

The WMS was nice because of the detailed diagnostics returned on
errors, and for the fact that we aren't filling up a directory with
image files.   On the other hand, the lack of ability to dynamically
control the symbology is becoming a show-stopper, and with CGI we can at
least throw it a URL containing terms like
"map_layername_class_expression=('[PROJECT_ID]'='89' ".

Similarly, will switching to a WFS service, or calling a WFS layer in
the WMS map file have any advantages?  I was looking into filter
encoding, but was getting lost by the distinction between returning
features and returning an image.  

At this point in our development cycle, it would be preferable to "drop
in" some functionality, rather than re-write a lot of code, but as I
said above, dynamic symbology is the driver.

Thanks,
Jim Cser