At 10:37 AM 12/18/2006, Chris Holmes wrote:
Hey Howard, couldn't figure out if there was a way to post comments on your blog, but thanks a ton for the feedback, and for taking the time to keep with it.
I don't allow comments on my weblog because I don't have the patience to deal with the spam and my weblogging system setup is old and quite poor (Plone + a weblog product + my own hackery).
The line feeds and the permissions on the binary release are super bad, and I'm really glad you brought up the issue. You're right, it's completely unacceptable, and we're working on it right now. It worked right in the past, but the problem is we transitioned build systems for 1.4.x. And no one had bothered to report this as a problem, so we had no idea, since our normal release manager tests on windows.
We haven't done extensive mac testing, as none of the core developers run mac, but we have a couple in the office that I'll try to spend some time with. The mac documentation was mostly a contribution from a user, I actually didn't even really know it was there.
Thanks for your feedback on documentation and the like. The transition to wiki actually was a boon to getting more information up there, but you've definitely pointed out that it at least needs serious cultivating and some serious thought on the information architecture. I've made a few attempts in the past, and generally keep a pretty type grip on letting it not fall prey to normal wiki practices, but I think it's been balkanized at this point. Probably surprising to you, but this is actually one of the first negative feedbacks I've gotten on the docs. But thinking on it, it's more in the past where I got really positive feedback, recently I've just gotten no feedback, and unfortunately with open source no news is not necessarily good news. So it's in need of an overhaul.
The wiki vs something else argument was a big one when MapServer was looking to revamp its website (me and my big mouth got me on the hook for that :). The upside of wikis is user-contributed documents (less work!). This is also the downside (not everyone agrees on organization or level of detail). MapServer had a primordial wiki that didn't have enough gardeners and didn't have enough "churn," and divergent and out-dated information started to creep in. To someone who knew their way around, this wasn't a problem at all. To someone downloading the code for the first time and having trouble getting going, this was extremely disorienting.
A few ways we try to alleviate some of this on the MapServer website, which has no shortage of conflicting documents, is to have the New Users page front and center and to have simple document associations that exist in the related box. Projects like MapServer and GeoServer have their own vocabulary, where terms are overloaded to have meaning within the context of the project. The new users doc gives people the gestalt of the project and the related links add a lot of context. These don't solve all of the problems of people being lost and confused, but I think it helps some.
As for packaging, the minimal approach generally works pretty well, except when it doesn't 
What were the tweaks, pops and pokes you had to do to get the bin working?
For the binary release, it was the FAQ of environment variables not being set. The stack trace didn't immediately whack me on the head and tell me so, however.
And what was the flavor of 'nothing' when you went to /geoserver after you dropped the war in to tomcat?
There was a bug in JIRA that fairly closely described an issue similar to mine in a 1.4. RC, but it was marked as worksforme. It was related to Tomcat's deployment of the war file. 1.3 worked like a dream in Tomcat, so I'm going with that for now.
If I have some time I'll try to get a specific mac packaging for geoserver-bin (all we do with the windows stuff is wrap geoserver-bin into a nice installer).
I don't think there's any need to do so at this point. It works, as long as your users RTFM 
Again, thanks for taking the time to actually write up the feedback, it's extremely helpful.
Thanks for being such a good sport about my post. I kind of regretted posting it, because venting my spleen publicly about my frustrations isn't all that helpful, and I didn't want to sound like I was just dumping on GeoServer (which I think I did). If I weren't intimately familiar with MapServer, I could probably use the same weblog post and just substitute all instances of "GeoServer" for "MapServer," and it wouldn't have to change that much to describe a newbie's situation on a underrepresented platform there either. The consistent mitigating factor is investing enough time and being persistent.
Howard