OA - Ubuntu
A blog about Ubuntu, mobile GIS and archaeology

OpenStreetMap + Launchpad? Not yet

Jul 01, 2010 by Joseph Reeves

At the end of last year I tried to encourage the use of OpenStreetMap within Launchpad rather than the existing Google Maps; there was a bug and a blueprint for people to comment on and I hoped that we might see Free Software embrace Free Data. 

 Since those bugs were first filed OpenStreetMap has come on leaps and bounds; it's saved lives across the world and has led at the forefront of a data revolution that everyone is getting into. The quality of the map data is improving rapidly, you can compare the coverage in your local area to Google's to see for yourself. Importantly, the barrier for entry to OpenStreetMap is incredibly low; if your street isn't on the map it's easily added.

 Sadly it seems like the logical marriage of Free Software and Free Data might not be; the blueprint has been marked obsolete and the bug closed with the note "We might reconsider this in the future when OSM provides all features that Google Maps does". I'm not sure what these features are, maps over https perhaps, but it is wrong to think that we need OSM to provide this. OpenStreetMap is a database of geograhic information that anyone can download and use as they like, even with the Google Maps API; we can provide required features.

 Launchpad is Free Software, but Google does not provide Free map data; Google provides map data at no cost. Would we host Launchpad servers on Windows Server if it was provided without cost? 

 A quality of Open Software that makes it better than closed equivalents is that it's Open. We all know this and need to apply the same thinking to data.



Disclosure: I am a Canonical-employed Launchpad developer, working primarily on the bug tracker.

Disclosure the second: I'm aware that you're ranting a bit here, but I'm going to say this anyway.

One thing that you don't mention here is where you think the resources to tackle this change should come from. Now, I don't know the ins-and-outs of Launchpad's Google Maps integration. What I do know was that developing it was a complex process; at the time it was written Google Maps was the only provider of the requisite mapping data that provided the features we needed.

The paid Launchpad development team is very, very busy. The part of that team that works on the areas that mapping touches is small and very, very, very busy. As Curtis Hovey of the LP Registry Team points out in a comment on the aforementioned bug:

The Registry team had a list of prioritised work, something must be dropped to undertake this issue. The Google Maps code was handed to us under the assumption it would be simple to integrate. It was not. We dedicated a developer to this issue for 1.5 releases. The cost was that team and project pages were not updated to the Launchpad 2.0 layout, nor have they been, because more urgent work was undertaken. There are new features and broken features that are more urgent then this issue--one of changing the implementation without adding for fixing Launchpad behaviour.

If someone from the community provided a drop-in replacement, I will review the code on my own time and work to get it merged into the tree. The behaviour is largely on the page, some deft updates to launchpad.js to support OSM/OpenLayer and GMap is the principle effort. Note that Launchpad is switching to YUI, and the script must be compatible. I will have to provide the code that toggle between GMap and the alternate maps code.

Now, yes, that comment was written over a year ago, but the amount of work that the registry team has to do hasn't diminished, and nor has the cost of switching from one mapping provider to another got cheaper.

On a final note, if you look at the bug you'll notice that it isn't closed and hasn't been for over a year. It's has a status of Triaged and a priority of Low, which accurately reflects where this bug ranks on the registry team's list of things to do at the moment.

I appreciate that it's vexing to see proprietary solutions being used in open source software, but I'd ask that you appreciate that the decision to use those solutions in the first place was motivated by pragmatic principles. If the cost of switching over to OSM were cheaper I've no doubt that we'd have either done it by now or would have reviewed a set of community-submitted branches that made the switch. Either way, the Launchpad team is not being nearly as stubborn on this as you make out in your post.

Posted by gmb on July 01, 2010 at 12:29 PM GMT+00:00 #

I like OpenStreetMap. I've even contributed some data to it, mapping some of the streets around my neighbourhood. I certainly have high hopes for it.

But it is not yet a viable replacement for Google Maps. My city - Osaka - is the second largest in Japan, and center of one of the largest urban areas in the world, and the map is still completely inadequate for even basic navigation. Major streets and throughfares are sometimes completely missing, and entire central neighbourhoods are still completely unmapped. Neighbouring Kobe, a city of a million and a half, is still largely unmapped.

It's not a complaint - the current situation is a vast improvement of only a year ago, and it's improving rapidly. But it's just not there for production use. Give it another couple years and this area will be in decent shape. But you just can't use this map reliably yet.

Posted by Janne on July 01, 2010 at 12:49 PM GMT+00:00 #

This is a good case when mapstaction http://www.mapstraction.com/ (assuming that it provides the required API's) would be the way to go as it allows the switching of mapping providers without having to re-write code.

Posted by Chris on July 01, 2010 at 03:07 PM GMT+00:00 #

Disclaimer: I am a Canonical employee, although not involved in Launchpad development.

Joseph, thank you for bringing attention to this again. Community interests shift and as I understand it the bug report has not been touched for a good 6 months now. Here's the link:

I am not sure why the blueprint was obsoleted but the bug remains open and as gmb indicates this will require more traction and effort from the community. Seeing as the bug shows "This affects 7 people", I guess it hasn't been circulated much, we need to get better at that.

Regarding missing or incomplete data, I will refer to the maps of Haiti, which were completed by the community in amazing accuracy and speed. Granted this is nowhere near comparable to such context but exposing this data in the first place is the only way it will improve to the level we need it to be.

I have seen this argument served many times about bugs I have hesitated reporting and features I have requested to be removed when buggy (a notable example being 3D GL screen savers). The answer from developers was: if we remove it, it won't ever get fixed. I tend to agree with that and I think it applies to this situation. But we need more than 30 people subscribed to the bug to make things happen.

Posted by FabianRodriguez on July 02, 2010 at 12:35 PM GMT+00:00 #

I tried to post a comment but it keeps being marked as SPAM!

Posted by FabianRodriguez on July 02, 2010 at 12:36 PM GMT+00:00 #

Post a Comment:
Comments are closed for this entry.