Source Code Isn't A StandardSource Code Isn't A Standard

Savio Rodrigues just wrote about a colleague testing a bunch of browsers based on the open source WebKit rendering engine. To his dismay, "no two are exactly the same". Or, as he put it, "A WebKit-based browser is, well, whatever the vendor wants it to be."

If you take a bunch of different applications that all implement WebKit, you would expect them to render the same things the same ways, but the evidence proves otherwise. My take-away from this little discovery could be summed up like so: Source code isn't a standard.

Source code is a building block, not a standard. It's something you turn into other things. A standard is something that stands above and apart from all of those things, a guideline for what that finished product ought to be like.

I get the impression many people who deal with open source expect source code to act like a standard, or serve as some kind of substitute for a standard. Ergo, something built on Linux should be cross-compatible with everything built in Linux. I see variations of this formula -- not always spoken out loud -- among people who see open source as a panacea.

The problem with using code as a standard is simple: it's too fluid. The minute you implement it in something, it's not the same code anymore. It almost always has to be changed to fit its container, as water changes to fit.

Savio argues that instances like this are a big reason why open source has to be paired with open standards. What you get out the other end has to pass muster with everyone, not just your own crew.

The problem is: how do you enforce an open standard in an environment where there are no real penalties for non-compliance? That's been bothering me for a lot longer than there's been a WebKit to only partly implement the standards in question.

information has published an in-depth report on Sun's future under Oracle. Download the report here (registration required).

Follow me and the rest of information on Twitter.