Wasted two hours today attempting to HTTP POST some content with a client certificate using curl on OSX Snow Leopard. It somehow would not show its cert to the Apache server.

In an act of desparation, I tried the exact same command with the exact same client certificate on Linux, and it worked.

So I downloaded MacPorts, built curl from there on OSX, and it works. No idea what happened, Google is of no help. I’m mostly posting this that others with my problem can find it.

Mind you, the NoSQL community still has a lot of work to do, years and years of work, InfoGrid and many other NoSQL technologies non-withstanding.

But I remember that when I first heard about what SQL is and what it does (particularly, what it can’t do), I thought: “this can’t be true. How many billions in revenue and market cap depend on that oddity?”. That was about when SQL was only about half as old as it is now… which makes this even scarier. (Try: no recursive queries. No abstract data types. No inheritance. No (meaningful) distributedness. No … <insert many other things here>. And how many thousands of lines would you like to write on object-relational mapping today? … )

So with that hat on, I’m proud to display this image as a bumper sticker, which comes from a presentation by Tim Anglade. May SQL never come near you If it does, run!

Disclaimer: I don’t build payroll systems for a living. If I did, I might think otherwise. But I think they have all been built, and the new stuff does require thinking much more along these lines.

What’s the next decade going to be like in technology?

I found myself pondering this a lot recently. It seems we are in for very revolutionary changes … like the becoming irrelevance of the PC. Or the move to NoSQL. Or all web apps being connected to each other, with RSS/Atom and OpenID being the first steps. Vendors, products, architectures, market dynamics will all be a lot different than we are used to.

Clearly worth pondering, or writing about it. Which not many people do. So I just started a new blog at:

upon2020.com

My focus will be the next decade, through 2020, thus the name, which of course is also a word play.

I am taking the risk that I might be terribly wrong with anything I might predict. It might be terribly embarrassing. But then, I hope to have a thought now and then that might spark some discussion, which is really all one can hope with on a blog.

So, enjoy! And disagree, otherwise, how should we all learn?

This blog will continue as before.

It was at the end of 2004 when I decided to start telling the world about this silly little idea I had had about a year before: give every person on the internet a URL that they could use to identify themselves to any website. Fully decentralized, no permission needed from anybody, under control of the user and so simple to implement and host, it could literally be everywhere.

This week the OpenID Foundation announced that now, exactly 5 years later, more than one billion identity URLs (now called OpenIDs) are operational on the internet. Not bad, I’d say. From 1 to a billion makes a compound annual growth rate of something like 6300%, over five years.

Time to compare the original vision with what it turned out to be. Well, some salient aspects of it anyway:

I did not run for the OpenID Foundation’s Board of Directors this year. I think I’m done there: I’m more of an inventor and innovator and entrepreneur than somebody excited about the daily grind of non-profit work of getting those billion OpenIDs used more every day, one day at a time.

Looking backwards, I think I need to be supremely amazed that this “silly” idea has had such amazingly powerful legs to walk that far. To be clear, if I hadn’t thought of it (and my wife Tammy hadn’t prototyped it), somebody else would have within a couple of years, most likely. And many, many people brought their ideas into the picture without which we would not have come to where we are. Thank you all, this is a story of collective barnraising. Success always has many fathers parents, and I mean that sincerely; in this case probably about a dozen. But still, it’s amazing to look back and trace a straight line over 5 years to the idea of the barn in the first place, and its basic architecture. Here it is, the barn, 5 years later, a billion strong. Not many times that anybody can claim to have had a hand in sparking something that became billions.

The jury is still out whether any meaningful money can be made around this. But I’m getting more optimistic: a billion is hard to ignore, in particular if all major players are on board, which they are. So going into 2010, I’m feeling like it’s time to do some serious business, and I think I know just where to start (contact me if you like)

So far, so good

Happy Holidays to you all!!

Just finished watching Kim Cameron’s talk at the recent Microsoft Professional Developers Conference. A bit of a surprise that talk of WS-* has largely disappeared in favor of much about REST.

But the most interesting part, for me, was at the end, when Gert Drapers (Principal Architect, Identity and Access Platform), gave a demo on future directions for Microsoft’s LDAP directory. Kim called it “two orders of magnitude simpler” (for the developer) than LDAP so far. The secret? Graph traversal!

Here’s a code fragment he showed on screen (I simplified it a bit to make my point):

Party me = directory.GetPartyByIdentityKey( ... );

IEnumerable<Party> managementChain = directory.GetRelatedParties( me, System.Identity.Kinds.Relationship.Manager )

// Find the first manager which is a expense approver
foreach( Party manager in managementChain ) {
    bool isApprover = (
        from roles in manager.ProcessRolesAre
        where roles.KindID == System.Identity.Kinds.ProcessRole.ExpenseApprover
        select roles.Party
    ).count() >= 1;
    if( isApprover ) {
         ...
    }
}

Here is how we would do it in InfoGrid:

Party me = meshBase.findMeshObjectByIdentifier( ... ).getTypedMeshObjectFacade( IdentitySubjectArea.PARTY );

MeshObjectSet managers = me.traverse( IdentitySubjectArea.ISMANAGEDBY.getSource() );
while( !managers.isEmpty() ) {
    Party manager = managers.getSingleElement().getTypedMeshObjectFacade( IdentitySubjectArea.PARTY );

    if( manager.getIsApprover().value() ) {
        ...
    }
    managers = manager.traverse( IdentitySubjectArea.ISMANAGEDBY.getSource() );
}

There are some minor differences in the API, because it appears that Microsoft’s is a special-purpose graph database with a built-in “directory” schema and a leaky SQL underneath, while InfoGrid’s supports any kind of model (aka schema). InfoGrid can also be run on top of either SQL or NoSQL engines and does not leak SQL. For this example, I made up a hypothetical model called IdentitySubjectArea, but that would a really easy one to define.

“Two orders of magnitude better” according to Kim? Of course, the world’s information is clearly structured more like a graph than LDAP and people seem to get around to that idea. Perhaps there are some interesting applications for InfoGrid as an enterprise directory … never thought of that one.