Recent diary entries
For the longest time it was claimed ODBL would better protect data than CC-by-SA in some jurisdictions, with the US being one of those.
However the opposite seems true, since the above claim was based on the premise that creating maps wasn't a creative enterprise.
The ODBL doesn't place a limit on what license produced works can be licensed as, they can be published as PD/CC0.
In any case unless the copyright license contains no derivative clauses people are then able to derive data from produced works and that derived data can be used to build a vectorised database.
There is one clause here where countries with database rights, when the data re-enters those countries the database right might re-apply, but this doesn't apply for countries like the US (or Australia for that matter).
Although I'm told that the above section of Database Directive in EU is untested in court, and I think some CC licenses already waive database rights and going into the future I believe creative commons plan to include this in more licenses.
One more point for those that might claim this would be difficult to do large scale, SVG files like the ones osm.org publishes are produced works, even though they aren't raster images, so converting to SVG and then back to map data would potentially be pretty trivial.
In other words CC-by-SA protects data better than ODBL, which is the complete opposite that we keep getting told.
The reason CC-by-SA will protect map data is copyright laws cover maps and the creative effort that goes into making them, it doesn't care how the map is stored, so while database protections are usually tricky this isn't the case when it comes to maps.
David Dean did a follow-up interview on ABC national radio this morning:
You can listen to the interview online and the interview last year:
Mapping a few things locally and noticed some things don't render, so I filed bugs about them:
Rendering waterway=drain areas, these are now very mappable anywhere Nearmap has made imagery available, but only ways currently render:
Rendering operator=* tags if name=* is missing, currently nothing is rendered but operator shows up in JOSM
Generic shop icons, I think it's important that something renders on the map, even if it's just the name and a generic icon for points of interest like shops, currently only a select subset of shops are rendering
Does anyone else have any thoughts on what else should be rendered?
There is a need for Object ID Permanence, currently this need is partially filled by using OSM table IDs, but these IDs are transitory, and can easily change for example when a node is converted to an area, merged with other nodes or even if there is similar nodes in an area.
However not all objects in the OSM DB need a unique ID, and some objects need multiple IDs, I've typed up a wiki page with some initial thoughts on the subject, and using UUIDs:
GUEST SPEAKER: David Dean on OpenStreetMap, a web 2.0, crowd sourcing, social networking development.
Come along on 5:30pm Tuesday 25 May to the first 2010 SITSIG meeting in the new venue, ACS Level 7, 333 Adelaide St (Theodore Building, between Creek and Wharf Sts http://osm.org/go/ueD3CAkwX-- ).
All welcome, free.
Shoaib Burq recently gave an indepth and interesting presentation at GA.gov.au on OSM and mapping of Haiti:
Due to all the images, the PDF is 42M in size, however the PDF has talking points to make it more interesting than just the presentation on slide share.
www.Nearmap.com released imagery covering the Albury/Wodonga area of Australia in the last 3 days, at which point there was about 70kB of compressed OSM data, as of today there has been a 7 fold increase of data to about 500kB.
I've been fixing up postcodes on and off for quite some time and for the last few days have been working toward completing Queensland and I'm pleased to have now finished the entire state.
There has been a number of people in Australia reporting that they have been stopped and questioned and/or had a gun pointed at them due to their activities associated with OSM. Same thing happens to Geocachers and a solution they came up with was to produce license like documents to help explain their presence.
So I went ahead and mocked this one up:
I've been photographing some of tourist route signs that use icons instead of numbers lately. I've also been playing with the trace feature in inkscape, which turns out to be surprisingly easy.
I made this one this evening:
Which I'm hoping can be rendered in future by people wanting localised highway shields.
I've spent a fair bit of time thinking about this and discussing it with Mike, and we think we have a suitable enhancement for osm2pgsql, we just need to tweak things a little to trigger some SQL queries after osm2pgsql has dealt with it's main task, and to null out the icon field if a road is updated so that new icons can be assigned if needed.
Hawke has also spent a fair bit of time updating the lookup page with suitable highway shields for the US:
I didn't hit this, but I helped find an answer for someone else having problems, there is a problem with Debian and possibly distros based on Debian where java will keep erroring and thinking it has no network connectivity, thankfully someone has already posted a solution to the JOSM bug tracker:
Nearmap is really a fascinating set of imagery, not only can you access the latest imagery, but you can also access past imagery which means if there is clouds obscuring your view on the latest imagery, you can use older imagery instead.
However to access this previously with the slippymap plugin in JOSM you had to go to all sorts of lengths using an apache redirection to do it.
I spent some time today cooking up a patch to work round this, now you can supply a regular url that uses x=,y=,z= parameters and JOSM will load it, which should be useful beyond nearmap, however your mileage may vary :)
Well the xmas holidays are here and I've been travelling about a bit mapping the odd street and even another miniature railway, while the railway doesn't render the name does which just looks weird:
I've submitted a feature request to get these tourism attractions rendered by the main OSM style sheet, which would be yet another POI only mapped and shown by OSM :)
In any case if you have small children I recommend checking out the mini railway in Casino, NSW it runs every sunday from 10 till 4 and only costs $4/ride, although there was only 1 person on the next trip so we all got a second trip thrown in :)
Oh and I quickly whipped round Casino afterwards and tried to get as much GPS data as possible, others were with me so I couldn't spend as much time on this as I would have liked.
Casino is one of those towns that most people just travel through, it's only by chance that I'm even in the area. It's not exactly a small town (by Australian standards that is) since there is 15,000 people live there, but it's just 1 of hundreds of such towns that just haven't been shown enough OSM love yet.
I've slowly been importing aussie postcodes and earlier today I pasted a personal mile stone of sorts, I managed to complete all postcodes for NSW which will most likely have the most postcodes of any state or territory of Australia.
Previous I've managed to import all postcodes for the ACT and the NT, so far so good 3 down only 5 to go.
I spent a bit of time tonight converting data on http://data.australia.gov.au to OSM files that can be loaded into JOSM and from there imported into OSM.
All up this should increase the number of POIs within Australia in OSMs DB shortly, for anyone interested there is a number of threads with information about the data on the talk-au mailing list.
Recently potlatch was updated to include the URL by default in potlatch options, this means you just need to click on settings and select it from the drop down rather than needing to manually enter a URL in the custom section.
In case you haven't seen the NearMap imagery yourself, the resolution and quality is better than Google sat imagery. The truly astounding thing is the imagery is only out by 1-3m from known locations and over time I'm sure NearMap will refine their processing to make this even more accurate.
Also we, the Australian OSM community, are currently working on show casing Carnarvon in Western Australia from NearMap imagery by mapping it out as much as possible. Carnarvon is a small town about 1000km from the nearest city.
I thought someone else would have made a diary entry about NearMap.com by now, in any case NearMap.com was mentioned on the talk-au mailing list by
James Andrewartha this past week.
They're currently operating in "Sneak Preview", but already online they have very high resolution imagery for Perth and surrounding areas, Adelaide, Brisbane and half of Sydney. As a bonus the imagery they have online is one to two months old, although on their website you can view older imagery as well.
For details on how to utilise their imagery in JOSM and potlatch head over to the wiki page documenting this:
I'm currently filling in some of the blanks around Gladstone, Qld from property boundary data. Some of Gladstone was already mapped out, however there is some largish areas that weren't mapped out and need surveying, if nothing else so street names and types can be mapped properly.
I went through last night and filled in a few blank areas using DCDB Qld data in the Sippy Downs area that couldn't be traced from sat imagery because it's out of date, possibly by up to 3 years.
That's the 10 million dollar question for the day.
There is a lot of roads that appear on paper, but were never built and this can be quite frustrating for people who go out to survey a road that just doesn't exist, it's even more frustrating if you consider you may have tried to survey the same road multiple times yourself!
Luckily this hasn't happened to me, but at least one person on the talk-au list claimed to have done this :)
The problem has been compounded recently by property boundary data becoming available for 1/4 of Australia, while this boundary data doesn't have roads it has clear outlines where roads and waterways and other similar features exist in a void between boundaries.
We (the talk-au list) have previously come up a list of suggestions of how to tag non-existent roads without copying the road into the OSM DB:
However with the current dataset we do have a dataset we can copy, it's just a matter of coming up with a useful tag to identify these roads.
So far the tags we've thought up are:
Can anyone think of anything better?