OLPC’s Paradigm Shift, and OOXML’s Flaws
I’d better post these before they get stale!
Chris Duckett writes Will OLPC Change Linux?
He examines the OLPC project’s ability to rethink a lot of entrenched computing paradigms, because it’s a new platform without legacy dependencies:
“The OLPC (One Laptop Per Child) project is in the enviable position at the moment of being able to change technologies and directions as they choose, with no rollouts and a lot of the work still to be done software wise. Free of the inertia that an existing product presents, OLPC can be fresh in its thinking and be a trend setter…
Another interesting choice was that of using application bundles, like OS X, and not a standard Linux packaging system. The reasons for this were that it is easy to install and delete, it is easy to share, there is no need for dependencies (because it is all in the bundle) and has an added benefit of removing the need for a centralized repository. If you think about children trying to use apt-get in the sub-Sahara, it makes sense to choose an application style that is decentralized and simple to use.”
Some very interesting thinking here!
Meanwhile, O’Reilly reports Groklaw Team Exposes Serious Flaws in Microsoft’s OOXML Specification. The original Groklaw analysis can be found here, carefully cataloging a long list of errors in the Microsoft spec. This analysis is an impressive piece of work.