"Randal L. Schwartz" wrote: > > >>>>> "Peter" == Peter Pimley <[EMAIL PROTECTED]> writes: > > Peter> ... he says, writing on his QWERTY keyboard > > <http://www.independent.org/tii/news/liebowitz_economist.html>
(Precis: the fable of the QWERTY keyboard - that it was designed to slow typists DOWN, and yet has remained in use - appears to be false. Leibowitz goes on to argue against "path-dependence" in other areas, apparently). *cough* 1. The real advantage of Dvorak appears to be comfort/health/safety, rather than sheer speed. (Speed is, once beyond the era of typing pools, less important anyway, I suspect, because more time is spent editing, opening, clicking, debugging etc than just typing stuff in). Many of those who have got their heads/hands around the Dvorak layout claim better comfort. Ask google, or me, & I'll dig out references... 2. The existance (or not) of path dependence is the subject of a virtual religious war in some quarters. Briefly, if markets can drive themselves into deadlocked situations, where they can't adopt a better technology (because everyone would need to switch in a co-ordinated fashion), there's a role for goverment intervention. *ducks enormous flock of approaching vested interests*. Do _not_ believe the hype. 3. If you don't believe that path-dependence happens significantly, you will probably have to conclude that people use Microsoft software because it is the best software concievable (: *puts cat amongst pigeons* 4. As a programmer, the idea of being stuck to doing things in an old, familiar, seemingly efficient way is a familiar one. It's possible to extricate yourself from that... eventually. Shearing layers, to enable stuff to be changed bit by bit, help... blah blah blah, design, structure, refactoring, blah blah blah ... *Swerves violently, leaving the smell of burnt rubber, and is back on topic* Re MySQL and "real" databases - correct me if I am wrong, but disable ref. integrity checking is not unheard of under "real" databases (Oracle, etc), in a production environment, for the sake of better performance. Repeatable-read locking is another nice theoretical feature for a database, but is often disabled for performance (again, correct me if wrong). For projects I've worked on, being able to (statically) verify ref integrity against a database (a fairly trivial schema-parse-and-do-some-joins job) would have been nearly as useful as a database which protects its integrity in real time. (*cough* so I should have done that, then, really ...) I've also seen projects where the heavyweight transaction support of a "real" database is a deadweight, nuisance and pain in the arse, as well as those where it's very useful. MySQL does joins, optimises them, and does them fast. This is collosally different from "a bunch of flat files"... Postgres may be a better bet, but if I made decisions based on someone with a reputation saying "it's what grownups do", I'd be smoking 20 a day & pushing a pram. Cheers Tim