(About EEPROM error checking/correction) More interesting (or more practical for many purposes) is a check and correction scheme which works on single objects instead of the whole memory. If you have the requirement to store and load dynamically (and frequently) during operation, you need to use a method which immediately discovers errors at the moment of transfer into the working set of data. Taking into account power losses during write often goes hand in hand with other "defensive" strategies. Unluckily there isn't too much to find on the net about transaction schemes and stable storages (at least not in tech detail, mostly stable storage is assumed as a basis to explain sg. different). My implementation is a (out-of-fashion?) block oriented, directory-less so-so file system; I wonder if there are people in here who have devised/implemented different systems. Any hints?
Mark