On Fri, 2007-01-05 at 10:21 -0800, Les wrote: > > Instead of prompting every > > time for whether or not you'd like to save or lose all your work, > > why don't programs have a default for how many revisions you'd > > like it to keep and always save all changes unless explicitly told > > to exit without saving? > Disk sizes are larger, but not infinite (at least for most users.) This > means that with large edits (think graphics and movies), you get disk > full quite rapidly. Yes, that's why the program should allow you to set the default for how many you want. You can base the choice on some ratio of the size of your disk (or wallet) vs. your tendency to make mistakes. > For professional users on large systems, this means > even larger disk arrays. Or confidence that they won't make mistakes, whichever is cheaper... > This is yet another case of one size doesn't fit all. But it's rarely used even where it would be cheap and appropriate. For example, text editors could easily incorporate a full revision history in an rcs file but you only see it done in wiki's where the rationale is probably "I'll have to back out someone else's mistakes" when in fact everybody makes some mistakes sometime. > Isn't one of Murphy's laws that computer data and programs will always > expand to overflow any existing storage medium and only at the most > inopportune times? Yes, but you have to balance that against Moore's law of exponentially increasing resources becoming available. Software design should let you pick an appropriate tradeoff between human and machine resources to consume and change that choice as dictated by economics. -- Les Mikesell lesmikesell@xxxxxxxxx
