Hugh Winkler holding forth on computing and the Web

Friday, April 27, 2007

The Penny Drops

It's enjoyable, and instructive, to watch the penny drop for venerable DCOMster/SOAPster Tim Ewald: I finally get REST. Wow.

Instructive, because coming from a strong RPC perspective, Tim illuminates the distributed application problem with slightly different shades. I like this graphical model:

Every communication protocol has a state machine. For some protocols they are very simple, for others they are more complex. When you implement a protocol via RPC, you build methods that modify the state of the communication. That state is maintained as a black box at the endpoint. Because the protocol state is hidden, it is easy to get things wrong. For instance, you might call Process before calling Init....The essence of REST is to make the states of the protocol explicit and addressible by URIs. The current state of the protocol state machine is represented by the URI you just operated on and the state representation you retrieved. You change state by operating on the URI of the state you're moving to, making that your new state. A state's representation includes the links (arcs in the graph) to the other states that you can move to from the current state.

Tuesday, April 17, 2007

Austin to Paris in 30 days

Since I'm traveling to Paris soon, thought I'd get directions from Google Maps. It's going to take 30 days, 9 hours. Notice item 28. (via Peter Flanagan).

Tuesday, April 10, 2007

Microsoft is Dead

Ha! I've been telling people Microsoft has become irrelevant. And now Paul Graham crystallizes the thought. Especially rich:
...I'm now surprised when I come across a computer running Windows. Nearly all the people we fund at Y Combinator use Apple laptops. It was the same in the audience at startup school. All the computer people use Macs or Linux now. Windows is for grandmas, like Macs used to be in the 90s. So not only does the desktop no longer matter, no one who cares about computers uses Microsoft's anyway.

An irascible colleague at a large software company used to say, "Hugh, you have to understand: XYZ isn't really a software company. It's an old folks home for software." XYZ had the same problem PG describes:
Microsoft's biggest weakness is that they still don't realize how much they suck. They still think they can write software in house. Maybe they can, by the standards of the desktop world. But that world ended a few years ago.

Tuesday, April 03, 2007

The cure is worse than the disease

This paper from Fortify makes the case that sending sensitive information using JSON exposes it to cross-site maliciousness. GMail sent your contact list down as JSON and evaled it. Turns out, any old site could do the same: just put a <script> tag referencing that contact list, and install some interceptor code that overloads setting e.g. the "email" property on any object: That enables the malicious code to see the values in the JSON.

Here are a couple of their proposed measures:

1. "Add the session cookie to the request as a parameter." Knee-slapper, that. See, the exploit only works because vulnerable sites put your identity into the cookie, and use a single URL for all users to download the object; the server uses the cookie to send you your personalized contact list. So the attacker just has to hardcode <script src="">. The paper proposes uniquifying the URL. Here's an idea: design your app so that each user's info is at a unique URL in the first place!

2. Send all legitimate requests for JSON data using HTTP POST! That way you know any GET requests are malicious ones from <script> tags. They do concede that "The use of GET for better performance is encouraged by Web application experts from Sun and elsewhere". There's no use for this measure if you use unique URLS, of course.

So yeah, this is a serious problem, but not for apps using best web architecture practices. Millions of web developers read papers like that and then crap all over the web.