Use of eeprom_update_byte instead of eeprom_write_byte

[leomil72]

Using eeprom_update_byte the code won't write on the EEPROM cell if it already contains the value that the user wants to write into it.

[SpiRaiL]

Thanks!

[mikaelpatel]

Actually I think it allows avoiding an erase cycle if the new value has the same "zero" bits and there are "one" bits in any place where the new value would require a change to "zero".

The erase cycle will force all bits to one (0xff) and then the new value is written.

Not clear if the eeprom_update_byte really does this as there is the below bug report:
http://savannah.nongnu.org/bugs/?38192

[matthijskooijman]

Seems that avr-libc only skips the entire write if the EEPROM byte is identical, it does not skip the erase when it is not strictly needed:
http://svn.savannah.nongnu.org/viewvc/trunk/avr-libc/libc/misc/eeupd_byte.S?root=avr-libc&view=markup

Switching to update does makes sense if only for the "do not write if the value already matches" (and the bug you linked will eventually be fixed, I'd expect). However, it seems that the current toolchain offers avr-libc 1.6.4 and the eeprom_update_byte function wasn't released until 1.6.7.

@leomil72, could you update your pullrequest (just commit --amend and force push) to only use update #if __AVR_LIBC_VERSION__ >= 10607?

[Chris--A]

The EEPROM library now provides an update method. EEPROM.write() still uses eeprom_write_byte.

EEPROM.put() uses the upadte() method to write its data.

[facchinm]

Fixed with bed78fd