Doing it completely properly requires that any DBMS storage *be under the control of* the CMS and versioning engine. That takes you into Zope or possibly even Frontier territory.

Frontier’s beautiful; shame the Linux port stalled.

Wanna start a development project?

You need the start date/end date versioning stuff, and the ability to see what the site will look like on a given date, using the *approved* versions of each element.

But you also need the ability to have different developer each pinch off their own version of the site, and see how that looks as they work on it, invisibly to both public viewers, and privileged viewers who are looking at a future (or, I might point out, past) dated view… only those people who log into an “SVN tagged” view of the site see those changes; this way you can salvo large changes site wide for testing.

All of this stuff lands in whatever piece of code does object-call dispatching, as that’s the only obvious centralized place to get it accomplished.

It also requires that you don’t short cut anything — notably, any context you choose not to place directly in the versioning system has to be stored in a versioned fashion as well; CSS and media files are the likely culprits.

WebGUI does a lot of this stuff, and I think Zope does as well. I don’t know about Django. One’s perl and the other two Python, so if your team is married to PHP, they’d be a hill to climb…

and copy that to the appropriate server to use as your static site-down page. You may need to grab the CSS manually…

(Captain Obvious, signing off

If your Apache configs are broken up correctly, then taking a piece of an internal site off line becomes merely a question of swapping a config file and doing a Graceful.

Often, if a problem is hard to solve, the best idea is to try to find a way to substitute a problem that’s easy to solve.

Not that symlinks aren’t clever, mind; but what happens if you have to take the whole machine down?

A couple of round-robin reverse proxies with synchronized config files make that trivial.

PS: Do you have a copy of TPOSANA on your shelf? Limoncelli put some great skullsweat into that book.

Shame you missed Doc; he’s really sharp.

I grew up in Boston; moved to Florida in 81, but various factors have conspired to bring WGBH back into my sites (and the local NPR affil thanks you; David Ives’ 35 year old plea for support just got me to send them money last week

Love behind the scenes blogs; I’ll probably keep an eye open here. Have you ever looked at WebGUI? Or, for that matter, Django?

The video session (opposite the How-To) was a great show-and-tell with some impressive video tools. One really cool concept that came from that session was from Richard Morgan (Twin Cities Public Television) who has set up a way to allow users to search closed captions for video which is just genius.

As an added note, I attended the “Think It, Build It” (I love that Daft Punk song, btw) session and really enjoyed it. It was one of the more refreshing experiences at the conference. We were actually asked to think (I know crazy, isn’t it?) rather then just shut up and listen. I’ll admit the developer in me wanted to not talk to people outside of Twitter and blog comments, but I did it and actually had fun.

It was an exercise in Design Thinking which boils down to taking time to think about your users before you even decide what you want to build. After you start to understand the user you can begin brainstorming ideas for applications/sites/etc that benefit them…not just things that we as developers/designers think would be fun to build. Smart.

And plus I got to geek out by attending a session with Merlin Mann…if you know who he is and are willing to admit it, you are definitely a geek.