OS X and I got Geoserved
Attempt #1 - Try to install it on OSX Server’s JBoss (45 minutes)
My machine meets all of the requirements of the Install Mac OSX document on the site. I tried following the scant document, even interpolating some necessary things like changing file permissions and ownership when needed, but I was never able to get it working this way.
Attempt #2 - Try to install it directly in a fresh Tomcat (45 minutes)
Next, I thought it my be prudent to try install in a fresh Tomcat instance, as this would likely be the most common configuration and installation of Tomcat. I got Tomcat working and I grabbed the geoserver.war file and dumped it into webapps. Restarted Tomcat and went to ./geoserver and got nothing.
Attempt #3 - Try to install the packaged binaries (geoserver-bin) (1 hour)
After converting the linefeeds of the startup scripts (dos linefeeds in a shell script that supposedly starts the server is inexcuseable) and following the documents for setting, java finally dies with a bunch of java.lang.NullPointerExceptions.
Update: After some more tweaks, pops, and pokes and synthesizing the Geoserver wiki into my neural network I was able to get this approach to work.
My Failures
- I’m impatient and I don’t like to read documents. Geoserver caters to me because there isn’t that much to read.
- I’m running OSX, which evidently isn’t a common configuration. Java’s java, and it is supposed to run everywhere though, right?
- I’m not a java developer, so I don’t have the experience to weather common trip-ups that a developer would have.
Geoserver’s Failures
- Wikis are horrible for project documentation. They are by nature divergent and the result is conflicting information. I would much rather have a single document that is precisely wrong once than four or five documents that are imprecisely wrong in different ways. As a naive user, I have no way to descriminate between the various bits of information and determine which are right, which are wrong, and which are current.
- Geoserver’s (this is probably just java in general) packaging is very spartan. It and the documentation assume I actually know what I am doing. The common use case of dumping things on a windows box appears be covered though, so maybe that covers 99% of the users, and I’m just in a very small minority.
You could argue that I’m an impatient, no-documentation-reading, know-nothing user who hasn’t spent enough time with the project and you would be right. These attributes still should prevent me from getting something going and seeing what it is about. Good packaging is hard. Good documentation is hard. Do people have exactly these same frustrations when they start out blind, naked, and dumb with MapServer too?
Update #2 I maybe just didn’t spend enough time and was too impatient, because after three or four hours I was able to get things to work. The problem of not having sexy or straight-forward enough documentation and packaging for open source software can make the barrier quite high. I know that most of the open source GIS projects I’m involved with continue to have this problem as well (my rhetorical question about MapServer notwithstanding). Starting again with an unfamiliar project with fresh eyes highlighted how frustrating this can be. I have no suggestions how to fix it though, as writing documentation, packaging software, and maintaining project websites is not egoboo-generating work. Maybe someday the firehose will point at these projects and these issues will all just go away 