Que sera, sera

So, my free-ride at the expense of Incisive Media is coming to an end - I used to host all my sites on their network (and with 100 MBit/sec access, it was pretty speedy :-), but since I've moved out of the country, they've decided to go for a more-managed hosting solution. Unfortunately, that means they'll be moving out of the co-location facility that they (and therefore I) have been in for a few years. Goodbye Level-3, it was good while it lasted...

Of course, it wasn't just a free ride, I pretty much engineered their web infrastructure

• 
  
• a template-driven website engine (now with over 150 sites running on it, originally specced for 5)
  
• a comprehensive permissions system (subscribe to a magazine, or for a period, or for N views, or N free views then pay-to-view) at the heart of the template engine.
  
• the database behind it all, tying everything together
  
• an XML-RPC interface to the various components in the system
  
• various feed systems (XML, FTP, SOAP) that could be generated and maintained by Incisive
  
• a payments management solution,
  
• back-end integration into their publishing systems,
  
• a message-board system
  
• a bulk-emailer with graphical (non-technical :-) reports
  
• a jobs-indexing system for Incisive's advertisers, allowing the advertisers to control the look and feel within an incisive-provided template.
  
• a full-text search index (I tried the MySQL one, but it was way too slow, mine can do queries over a million pages in ~1/10th of a second, orders of magnitude faster than MySQL on the same hardware)
  
• integration of search into other components (eg: mail-merge, the editor does a search, and these form most of the body of the templated email, together with his comments intermingled in the prose)
  
• dynamic templating, so search results can be presented differently on different sites (for example)
  
• PHP modules, evaluated inline to provide dynamic content (this is how the different search results was done :-)
  
• a back-end management system that presented different interfaces depending on the technical level of the user (as decided by Incisive :-)
  
• a dynamic firewall - if the webserver didn't like its input, it could add the current IP address to the list of those blocked by the machine's firewall. Since it was machine-generated, it stored a reason why the firewalling had happened as well, which came in handy when irate businessmen 'phoned up to ask why they couldn't see the site [grin]
  

One of the things I'm reasonably proud of is that the system was both modular and scalable - it started off as a simple templating system, and grew, (and grew...), but because I kept providing general-case solutions to specific problems, it scaled really well - to this day you can go to (eg: risknews.net, view the source, scroll down, and see the stats for the page in a comment at the bottom :-)

So, I've moved the blog over to RapidWeaver. Up until this point I've been using iWeb, but I was never really that happy with iWeb - it's a bit too limited for me.

I had actually been developing a web-creation system (and maybe I'll carry on) called 'Kaligraphy'. The idea was to take iWeb and make it more powerful, to make an application that was very integrated with a PHP engine on the web-server. The OSX-side was going to be just a layout-creation tool, with the text being flowed via a content-management system written in PHP.

I've written a content-management system before ... I wrote the one powering Incisive Media's websites, at last count there were over a hundred thousand webpages being served from that system. I wrote a full-text indexing system to run over it that was an order of magnitude faster than the MySQL built-in text-index. I wrote a mass-mailing engine (*) that could key off that search-engine to produce dynamic results and send a million or so emails per day with full read/response tracking. I wrote modules that allowed subscriptions by product, by page-count, by time, etc. What I am trying to say is that I have some experience in dynamic websites... I was hoping to bring that experience to bear while creating a graphical front-end on the Mac rather than the form-based front-end that we have right now.

To bring this post (kicking and screaming) back on topic - I've moved to RapidWeaver mainly because I have another 'big' project underway , and the fact that it has a plugin architecture. That ought to give me sufficient flexibility for what I want to do, without having to write a complete application to do it. The new idea is fantastic - even revolutionary :-) so I can't see myself finishing Kaligraphy in preference... So much to do, so little time...

I was considering going with SandVox, but with the 'MacHeist' offer available, I thought I'd try RapidWeaver instead. I've seen the pro's and cons argued back and forth (and to be honest, I'm not sure I'd have gone with the offer if I was a developer of one of the products), but from an end-user's perspective, I think it's a good-enough deal to go for. Personally it was the website-layout tool and one of the games (enigmo2) that made it worthwhile. I guess everyone will be different...

I guess the other point is that some poor deluded folk have actually subscribed to the RSS feed on here, and they're going to need to resubscribe to the new feed if they want to keep up-to-date.

(*) Incisive are good-guys - you only get mailed if you've genuinely signed up for something, and every mail contains a link that will actually unsubscribe you, rather than add your email to a "verified" list as some people do.

There’s been a fair amount of talk over how unsuitable a bunch of touch-sensitive areas down the sides of a screen would be, to replace the traditional iPod clickwheel.

I’m not-so-sure that that’s a valid criticism - quite apart from the patent clearly showing virtual clickwheels on the main display area, it ought to be possible to use a vertical strip of touch-sensitive areas as an input mechanism

The crucial part is the “infinite length” of a circular motion - you can describe a really-rapid motion with your thumb/finger and it’s always going in the same direction (clockwise or counter-clockwise). My point is, though, that a circle is just a particular style of ellipse. Imagine squeezing an ellipse so it’s very thin and very tall - you could still run your thumb up and down that ellipse by making your thumb go in a circular motion, but having the sensors on a line.

Physicists call this the vertical component of the motion - it’s still the same, no matter how the horizontal component is changed. Try it: hold a pen in your non-favoured hand, and with your favoured hand, put your index finger at the top of the pen then move your thumb in a circle. Et voila, your thumb slides across the pen surface, and sensors there (in a line) could detect that cyclic (it’s no longer circular, but it is cyclic) motion, and do the same as a clickwheel would do.

It seems obvious once you’ve thought of it - and I wonder why no-one else has come out with anything like it before Apple filed this patent. Perhaps it’s not-so-obvious after all...