One major problem with tech journalism is that most of it is based on regurgitating marketing materials and press releases. Microsoft for decades has spent huge truckloads of cash on marketing campaigns to burnish their image, from traditional advertising to astroturf comment campaigns across tech journalism boards. Being a prime source of tech advertising dollars, it was rare to find a PC magazine that was willing to bite the hand that fed it and independent commentary was flooded with astroturf. (The Natalie Portman/grits trolls weren't the only reason Slashdot introduced its moderation system.)

So as expected the comments ...

So with the upgrade to php 5.3.4, I decided to take the plunge and move to php-fpm from the standard php-fcgi. It took a few tries to get the move to an external fcgi server worked out, so you may have noticed a little flakiness this morning. But everything seems to be working now, and we'll see what the stability is like in the coming days.

Maybe instead of polluting the HTML5 namespace with a tag soup of confusing semi-semantic tags, Google should have taken their data on common id and class names and worked with the Semantic Web community and the Accessibility community to extend the WAI-ARIA roles in to a more comprehensive vocabulary on the structure of web pages. To be machine readable, you absolutely don't need to accommodate the average structure of your average WordPress/Blogger page, while simultaneously handcuffing designers and developers who want to structure their pages in experimental ways.

So I've installed clang from FreeBSD ports and updated sources to 8.2-Prerelease. I couldn't get the buildworld to succeed with clang, but buildkernel did. So this is the first time I'm running a non-gcc kernel on the server. Lots of warning messages during compilation concerning ignored compiler flags, but we'll see if it makes any difference to actual functioning of the kernel.

Inspired by Microsoft's rush to claim to lead in the rush to HTML5 and Google's self-congratulatory online book to show off Chrome's HTML5 capabilties, I decided to check out how they were handling HTML4's red-headed step-child: the OBJECT tag. Most web developers understand that support for the OBJECT tag is broken, but most of them don't really care as it's seen as only a way to include Flash or other proprietary or antequated binary data like Java applets.

Actually the OBJECT tag should be an easy way for web programmers to include any binary ...

Lately the commute has provided more time for catching up on podcasts. One of these is Jeffrey Zeldman's Big Web Show. For those not familiar with Zeldman, he's been one of the forces behind Web Standards since the days that Microsoft was rationalising their decisions to maintain IE6's "beloved" box model.

His guests are some of the top folks in the web industry. So while listening to Episode 15 with Tara Hunt. It's a long discussion covering many topics, but my jaw dropped during a section discussing her new startup, Shwowp. They were discussing metadata and ...

The FreeBSD ports tree has both spidermonkey (lang/spidermonkey) and couchdb (databases/couchdb), but they are a little outdated. While trying to build riak on FreeBSD, I needed to get spidermonkey 1.8.0rc1 to build. So after examining the patches added for 1.7.0 and playing for a while, I got 1.8.0rc1 to build. Decided I'd share the port with the world, so here's the port dir for Spidermonkey 1.8.0rc1. Create a /usr/ports/local and unpack it there and build.

After that I needed to rebuild CouchDB. Checking the Couch homepage ...