Virtual Dumb Terminals

Dumb made smarter

When I was studying Computer Science at university we used Digital UNIX on a server and rooms full of dumb terminals connecting to it. It meant that the terminals simply showed the display and passed the user’s input onto the server, hence the ‘dumb’ title.

The world of computers has moved almost exclusively into the opposite model of stand alone computers or at least ’smart’ terminals that use a server but process most work locally. The latest move into ‘cloud’ computing is not a move back towards the dumb terminal model because the actual processing of data is still done on the local machine.
But there could be a move back towards this model, as virtualisation becomes more and more pervasive.

My idea is to take a virtualisation server and let it dynamically load the ‘terminal’ computer that runs the virtual machine with whatever the terminal can handle. The server would monitor things such as the RAM and CPU load on the terminal, and if resources are running low perform more calculations on the server end and simply send the result to the terminal, which would require less resources.

This would mean the terminal could always run fast and light, and the server could offload as much as possible to the terminal, but bail it out when it needed some extra muscle.  There are many advantages to virtual computing such as load balancing, easier software upgrades, more choice of platform and so on.  This approach could make it much easier to upgrade hardware as well though, as a new server could result in better performance across all it’s terminals just as if the terminals has been replaced instead.  Older hardware would be able to run more advanced software (albeit with an increased load on the server over newer, more capable hardware) and everything from smartphones to main frames could be supported centrally.

Search vs URL

Something that happens to me quite a lot is when I’m trying to type into the search box in Safari I accidentally type into the Address field, and end up going to latest-news-on-apple.com which of course drops out with an error.  But when you think about it URLs almost never go in the search field and search terms never go in the Address field.  Why are they not the same bar with some intelligent URL parsing technology to separate the two?

I know some browsers use a single letter keyword to indicate a search rather than a URL but surely when Google Calendar can figure out when to put an appointment just from the natural language then a browser can figure out what’s a URL and what’s not?  A keyword search is nice – but it could be so much more.

And the possibilities go on – how about some AI that intelligently figures out trends in image searches and redirects likely image searches from web search?  I suspect Google already does something like this in their search algorithm by placing likely hits at the top.

Note to Apple, Microsoft et al:  Feel free to steal this idea and use it without credit to me – as long as we get better browsers out of it!

Sneaky BBC

I’ve been enjoying watching catchup TV on the BBC online iPlayer, shows like Top Gear, Never mind the Buzzcocks and Lead Balloon.  Since I have a Mac of course I can’t download shows like my WinXP friends can.  But I have a secret weapon – Parallels!  So after installing the downloadable iPlayer manager in WinXP running in Parallels I was all set to download my favourite shows.  But no.  Sneaky, sneaky BBC sets the default downloaded store to 0 Mb.  And when you try to change it, it pings back to 0Mb before you can save the settings!  Very sneaky BBC!  Give us a way to download shows so you seem like you’re giving us some download loving, then cripple it to make sure no-one actually does!

Mac OS X Uninstaller

Having just read John Siracusa’s excellent (as always) review of Leopard, I got thinking about one gaping hole in Mac OS X. An uninstall application/framework.

Apple’s Installer application and underlying framework is right up there, it’s easy to use, easy to create installer packages (and metapackages) for, easy to use and parse via the command line. But where is it’s uninstalling counterpart?

Apple have always emphasized single shot uninstallations, as in, you delete the .app package for an application and it’s done. Simple. Unfortunately applications are not just the contents of the .app package. They create preferences files in the Users or System Library folders, amongst other things. Especially with the new application signing feature of Leopard, would it be that hard for Apple to keep track of files created by a specific application using the fsevents framework that keeps track of all filesystem changes (as leveraged by Spotlight and Time Machine). Then when an applications is deleted, the system could offer to clean up other files that had been created by the application.

Of course, this could probably be accomplished without either application signing or fsevents as long as applications stick to naming conventions so associated files can be found. But using these advanced frameworks could lead to a scenario where you delete an application, and Mac OS X pops up and asks you if you’d like to remove preferences files, widgets, preference panes and even documents created with the application!