Mikel was inspired to put free data on his iPhone - so he figured out how to
use OpenStreetMap on his iPhone Maps Application.

We tried a number of methods - it should be as simple as changing the tile url that the Maps application queries. But initial “peeking” under the hood didn’t reveal where GMMServerURL is set. In the meantime - Mikel ended up finding the tile cache sqlite database (you don’t want to continually download tiles) and stuffing it with OSM tiles instead. So as far as the Maps application is concerned, its serving cached tiles. Pretty slick.

As Mikel points out, there are some issues to resolve - tile sizes are different on the iPhone, so the zoom levels are off. That would be easy to verify by loading a KML file with markers.

It’s also a slightly sweet answer to Google’s Cease & Desist to MGMaps for using the tiles. Mikel’s “hack” shouldn’t violate anything, he’s just be very considerate to his data provider by pre-caching tiles that have amenable TOS.

I started to write this blog post a couple of weeks ago, but forgot to push “publish” - here it is now

Bad day and a half for the web. Both Skype and DreamHost (and more importantly DreamHost’s DNS) were down last night and today until just now awhile ago. The effect was that any of my sites appeared down and were unreachable, and again more importantly, no email sent to me would have gotten through. (and here I was amazed that by 11AM I didn’t have a single email!)

So if you sent me an email this morning - I may not have received it. They’re slowly trickling in - but if you don’t get a response from me, the maybe try resending it, or bonking me on skype/im/twitter/jaiku/plazes/facebook/et al.

This morning, I tried to login into Basecamp - and I use OpenID, which is rerouted from a .highearthorbit.com URL. So obviously that didn’t work. You can instead request a temporary account - but that can’t get through since the DNS doesn’t work. So - another single point of failure in my system.

Broader problem of single points of failure

While this was a single incident, that probably won’t result in (much) lost information, it does point to a broader problem. I have a single email server, and a single service for DNS. If either of these goes down, then I have an unknown state of email - and more than likely anyone sending me email will get either: a bounce-back that the mail was undeliverable, or worse, nothing - and they’ll be left thinking I just didn’t respond or bother reading their email.

No one uses “email receipts” anymore - and in general I don’t like them. I don’t want to share when I’ve opened your email, because then you get another email that I’ve read it, flooding your inbox.

However, I would be able to like to verify that a message was successfully received in some way - either just to the server, or to the client application. I don’t care if it’s been opened or not, I just want to know it got to wherever it needed to go.

I’m thinking I need a http://andrew_status.net, or even a more useful http://andrew.turner.mystatus.net - maybe even tied to my OpenID. There would also be a mystatus3 and mystatus4 hosted on other networks and infrastructures with paralleled information.

On it I would want to, at minimal, post any problems that affect my general communication: Andrew’s Email isn’t working (1 day), or Andrew sick, or On Vacation. No more of this auto-reply to emails that you’re on vacation.

I live digitally - almost all my contacts are online, and more importantly the ones online can’t verify my actual status. I can’t “tell” them my email is out. I don’t have an office or assistant to answer calls or emails with “Andrew’s Sick, he’ll be back in on Thursday”. Instead I need a MyStatus.com, it’s like a HeartBeat for PeopleApps.

I just recently finished developing another website application with a mobile client (will post about it shortly) and ran into the common annoyance of how to deal with logging into your account from your mobile.

For Mapufacture Mobile we require the user to login once, and after that we recognize their handset and don’t require login. This new site supports OpenID (coming to Mapufacture soon too), which isn’t really an option on the mobile handset. The client is through uLocate’s WHERE platform, so sharing a session cookie from a browser isn’t an option.

Also add on to that the similar problem of having to remember dozens of usernames and passwords for different services, while you’re out and about, and have a limited screen & keypad.

This seems to be exactly what OpenID was made for. I should be able to register my device to OpenID (phone number, SMS verification, simple web client) and from then on, have a single sign-in on my mobile that then authenticates any service on that connection.

OpenID Personal Directory

Not only that, but I should be able to log into OpenID Mobile and have a bookmarked list of these services. So once I login, I can pull up my blog posting, twitter, email, phonebook, etc.

I haven’t entirely thought through the mechanics. The concept is simple enough for web-capable phones using WebKit (iPhone) or Opera or whatever. However, how do you provide this on devices or for users that don’t have a web browser - and also enable the entire phone, not just web apps or WAP sites as Laurent points out.

Perhaps this is where OAuth comes in. OpenID servers should provide OAuth capability. This way anyone can request a validation token that works with other OAuth or OpenID sites from my actual OpenID provider.

Today Google gave a couple of presentations on the current state of KML to the OGC Technical Committee meeting in Boulder, Colorado. The purpose of these particular presentations were to put forth KML2.2 as an OGC best practices. KML still belongs to Google, but is in the process of moving to belong to the OGC and therefore be an open standard.

The big ‘announcement’ was that Google will be releasing an open-source KML library in C++ that implements and tracks the standard as it progresses. By providing a reference library it allows developers to more easily keep up to date with KML without having to maintain their own library and track standards changes.

Assumedly bindings could be written to make use of libkml in other languages like Ruby or Python.

This is smart for Google, and any publisher/consumer, as it helps ensure that clients and servers are properly using KML and there aren’t various mixed versions beings published.

The library is expected to be available the first quarter of 2008 - and available under an as yet to be determined open-source license.

Map of aggregated feeds from my life & travels: