"For what it's worth some of those blog postings are probably from countries where major highways still need to be mapped. Some people just aren't interested in mapping big things with little relevance to their personal lives and prefer to map things closer to them."

I'm sure we all do that to some extent. There's a certain amount of satisfaction to seeing the details of your own neighborhood show up on line. I rarely drive a car and am mostly out on a bicycle, so the extent of my ability to map is geographically limited by my strength and undurance.

"Eventually your region will be full of active mappers."

I hope you're right. At the moment I'm the only one within a few hundred K doing any active data gathering at all. There were more, but they seem to have dropped out.

As I've said elsewhere, the current OSM slippy map rendering doesn't do much for those of us in sparsely populated rural areas. I know one mapper who gave up for this very reason.

I have used GT Rider stuff in the past. I did a Laos road trip, pre-GPS, earlier this year based largely on information from their web site.

http://www.mgnewman.com/Travel/show.php?mode=index&show=Lao0902

But, they are a semi-commercial operation, so I doubt they'd be interested in OSM. I'll send them a note, though.

The map I use on my own GPS is from this place http://rotweilermaps.com/. Also semi-commercial

Great news emj. I started out trying to do this myself, but it's a huge task. I actually have mapnik running, both on my Mac and on a linux virtual machine.

I understand the problem of deciding when to use the countryside style. What are your thoughts, so far?

What I'm doing at the moment is looking at the osm.xml stylesheet which I got from here:

http://johnnydobbinscom.s3.amazonaws.com/osm.xml

I need to see how to revise that so that it does what I want with POI's at low zoom levels.

BTW, I got some good hints from this site: http://johnnydobbins.com/2009/05/31/mapnik-ogcserver-setup-from-scratch/

Slow going, but I'm getting there and learning a lot.

Thanks Vclaw. Great stuff. I've now got Mapnik running on my Mac, but the information in that tutorial is quite good and detailed.

The only thing I really want to produce is maps that show POIs at a lower zoom levels than the OSM default. This makes great sense for sparsely populated urban areas. We have lots of areas like that in Thailand.

Unfortunately, the CloudMade customized style sheets do not let you make POI's visible at lower zoom levels. They do let you go the other way: make the zoom level at which POI's are visible higher than the OSM default.

I am now in the process of teaching myself Python and Mapnik. So far I've been able to use Mapnik to generate customs maps with POIs as I want them. Unfortunately, learning Mapnik XML formats that let you customize maps is no easy task. For example, I still haven't been able to figure out how to use TextSymbolizer to draw text on the map. But, for a silly old fart I'm doing OK so far. But, I've spent the better part of three days just getting this far.

However, it is complete folly to expect ordinary people without substantial command line computer experience to actually install and use Mapnik. It ain't easy.

Turns out that the problem is that boost ends up getting linked to the wrong version of Python. No worries, it's easy to fix with this simple little command:

sudo install_name_tool -change /System/Library/Frameworks/Python.framework/Versions/2.6/Python /opt/local/Library/Frameworks/Python.framework/Versions/2.6/Python /opt/local/lib/libboost_python-mt.dylib

Now Mapnik is working on my Mac….

Since I couldn't make heads or tails of the mapnik wiki:

http://trac.mapnik.org/wiki/MacPythonUpgradeIssues

Which contains gems like:

"Mapnik will build just fine with your new python installation will not complain but when importing you get the "Fatal Python error: Interpreter not initialized (version mismatch?)". Check to see what your _mapnik.so is linking against."

and

"If, like the above command result, you see that your mapnik module it linked against 'opt' (the macports dir), then either remove boost from macports or switch to the macports version of python. Even if you pointed the SCons path variables at a different version of boost, the SCons linker found the macports version because it was on the library path of another one of your dependencies (In the above example, the JPEG_LIBS)."

I've decided to completely uninstall everything I installed from Macports and try again.

I hope Zartbitter reads this as it is at his invitation that I'm tilting at the Mapnik windmill….

So, I've searched around the mapnik wiki and found my exact problem. But, this is the….

----
Solution: In general you must rebuild Mapnik to link against what ever version of Python the boost python library linked against, or rebuild boost from source which will likely link boost python against your system version of Python.
The otool command is how you find out what is linked against what library. In the following example, Boost python and Mapnik are both linked against the Python25 Apple installed version of Python on 10.5 (which works):
——

This is all over my head. Once again the geek-kings triumph over the map-serfs….

Thanks for your comments, Daveemtb. I think you're right that there is a definite division between the geek-kings who write code and the map-serfs who gather data for them. I believe, protestations to the contrary, that the geeks look at this as "their" project. As a result, they don't take suggestions kindly:

"Telling us what colour you think the bikeshed should be doesn't get it any nearer to being built."

"Feel free to generate your own map…."

It's definitely off-putting and doesn't do anything to attract people to the project. And, as you noted, the frustration causes many of us to lose enthusiasm for what could be a worthwhile project.

Frankly, I don't have much optimism that things will change in a way that makes me comfortable with continuing to contribute.

Earlier in this diary entry I mentioned a friend who had given up contributing to OSM. Here's what he wrote to me recently:

"For a '60 year old fixed-income retired geezer living in rural Thailand' you express yourself quite articulately. Seems like a lot of people see the same shortcomings in OSM that you do. What to do? Well, first: Don't take it all too seriously. I think I am getting better at sometimes -- perhaps often -- simply saying forget it. It's not worth it and there are so many other worthwhile things to do. Not that I'm suggesting you/we should give up the mapmaking project. Just that we see if we (well you, really) can find a more productive direction to achieve the goal. Enough."

"Telling us what colour you think the bikeshed should be doesn't get it any nearer to being built."

I wasn't telling anyone anything. This is not a forum or a project mailing list. It's a diary. I'm recording my personal thoughts. If others choose to comment on them, fine.

Our product would probably have a lot more volunteers if our product was easier to use and had a friendlier and simpler face. Frankly, I think that the mock-up pages that CloudMade came up with are still too complex, but Fp4 is probably the best of the lot.

Perhaps when I'm a little more confident I'll start participating in the mailing lists. Until then I plan to continue to publish my thoughts here in my Diary….

If I could do any of those things I suggested, I would. About the only thing I can do is contribute data and do some mapping, which I do for many hours each and every week.

I'm sorry you didn't like my mock-up. I did it in about 20 minutes as a quick and dirty exercise to clean up what I think is an overly cluttered homepage.

This rest of today I've spent trying to install mapnik on my Mac. Sadly, the existing instructions:

http://trac.mapnik.org/wiki/MacInstallation

are both out of date and, in places wrong.

After eight hours of trying, I'm still no closer than I was this morning.

By the way, if a user of your product makes a suggestion or request, the best answer is NOT: "If you don't like it, do it yourself."

Kind of reminds me of the right-winger bumper stickers of the 1960's: "America - Love It or Leave It". Is that really the impression you want to leave?

I have a slightly different take on it…

One of the reasons Google gained market share on Yahoo! so quickly was because Google's user interface was clean, uncluttered and unambiguous. It did one thing and it did it well in a simple, intuitive way. OSM's home page should strive for the same thing.

Here's a quick-and-dirty mockup:

http://www.mgnewman.com/sz/osm_mockup.jpg

One of the reasons Google gained market share on Yahoo! so quickly was because Google's user interface was clean, uncluttered and unambiguous. It did one thing and it did it well in a simple, intuitive way. OSM's home page should strive for the same thing.

Here's a quick-and-dirty mockup:

http://www.mgnewman.com/sz/osm_mockup.jpg

Thanks to all for the constructive comments.

Collaborative community projects which depend on armies of volunteers can only succeed if the information they gather is accessible to the masses in a way that inspires even more volunteers. Think of Wikipedia. It has been wildly successful because anyone with a web browser can use it. The interface is clean and intuitive. It's all right there. But, how successful would it be if each search only retuned a summary of the article and to get the entire article each user had to compile and run a rendering engine on their own machine?

Look, I'm a 60 year old fixed-income retired geezer living in rural Thailand. The chances of me being able to compile and run Mapnik on my little Mac Mini are zilch. And, even if I were successful in rendering tiles for my part of the world, I'm not going to be able to serve them up to all comers.

I started working on this project with high hopes. Those hopes have mostly been dashed. I'll probably continue, but not with the enthusiasm and vigor that I had in the beginning. The rewards, in terms of contributing to the public good, are just insufficient. That said, I've learned a lot. Always a good thing when you're in your declining years.

"Feel free to generate your own map showing POIs at other zoom levels using the data."

Certainly, if I could, I would. I don't have the slightest idea how to go about such a thing.

And, believe me, with an arrogant attitude like that you're just going to chase people away from this project, not attract them.

For example:

A friend and I worked hard to map this area of Thailand:

osm.org/go/4ZA9uwr-

It's an ampoer (county) in Nakhon Ratchasima province. It's adjacent to Thap Lan National Park and contains dozens of funky little resorts and restaurants. We decided it would be a decent public service to map the area and mark POIs for all the struggling little businesses in there.

After mapping the roads and starting to locate various POIs we realized that our effort probably wasn't worth it. In order for the map to be useful you need to view it at the zoom level in the URL above, so that you can see that national park and the main highway. But, in order to see any of the POIs you have to zoom in so far that you have no idea where you are in relation to known landmarks.

Bottom line: my friend decided not to bother any more. Result: we've lost the only data gatherer we had in that part of Thailand.

Why not render each tile based on the number of objects present? That way you can render each tile just once using a density-appropriate style sheet.

I wish I understood why it was so complex. CloudMade allows you to customize style sheets, which seems to show how easy it can be. But, CloudMade customizations do not include displaying POI icons at lower zoom levels. In other words you have to be having a detailed look at the VB cans in the Todd River before you discover there's a pub in town….

It would be very nice if the CloudMade style editor had more flexibility in terms of the visibility of objects at various zoom levels. While it might make sense to limit the visibility of, say, hotels to zoom levels 16 and above in dense urban areas, this restriction is not appropriate in the countryside where accommodation is scarce. I can understand wanting to limit clutter in cities, but outside of cosmopolitan areas showing more POI's makes the map both more interesting and more useful.