Print This Post

GML “Sucks”

While perusing some other blogs I came across the following:

geek post–GML sucks

Posted on 2004.10.11 at 20:36

By now, I should be able to do something like:

1
2
3
4
5
6
7
8
  <map region="earth">
    <feature borders="political"/>
    <dataset xsrc="" mce_src=""/>
    <dataset name="lj community location" xhref="http://www.livejournal.com/userinfo.bml?user=conlangs">
      <city name="New York" parents="New York,United States" label="ouwiyaru"/>
          …
  </dataset>
  </map>

and be able to see what it looks like in a regular web browser. Then people wouldn’t have to have Map servers. News sites, Map Quest, blogs, etc., would send that data rather than an image — at least it would be possible to do so.

All progress seems to be focusing on server implementations rather than client implementations (RSS, maps, etc). New standards are comprehensive but do not lend themselves to simple, readable composition. Client implementations are concentrated in two to four unwieldy projects (mozilla, IE, etc). Microsoft is partly to blame because the only easy way to make universal apps is to push web-based IE-supported tags/APIs from the server.

Instead, plug-ins should enhance the experience, but not be essential; there should be a framework so that the plug-in is unnecessary-with a default java link or by building up with primitives (SVG?). This would semi-cripple the interface and/or slow down generation, but then the fallback would be on the client rather than on endless server encoded exceptions for each browser, etc.

The symantec-web people are also misguided. If Google has taught us one thing, it’s that a website cannot be trusted to describe itself. If clients have the capability to download and comprehend simple marked data, then people will publish and use it. It’s really that simple.

I thought this deserved some sort of response since I am sure that there are many people with similar views. The contention appears to be that any browser should understand some sort of map grammar and be able to render it. Of course this puts a rather big load on such a browser. For geometric objects (points, lines, polylines, polygons) this is a doable task, and one already served by VML in IE and via SVG in various plugins. After that the going gets rather tough.

Our viewer “ouwiyaru” wants the browser to understand some specific terms like “political” – but this is one of a possible list of maybe tens or even hundreds of thousands of such feature types. Perhaps the browser could just understand some limited, popular subset?

As for plug-ins vs native browser functionality – it is best if the browser provides foundation capabilities – like graphics (e.g. SVG, VML) and styling functionality (e.g. XSLT) – then you can do what you want – and easily “market” it to the rest of us.