java - Why use random numbers for serial version UID? -

java - Why use random numbers for serial version UID? -

this question has reply here:

why generate long serialversionuid instead of simple 1l? 9 answers

edit: random mean big computed number no semantic meaning developers

when implementing serializable interface best practice , of import specify serial version uid. in numerous places, see random numbers beingness used. e.g.

effective java (2nd edition) pg 312:

private static final long serialversionuid = 234098243823485285l;

from string class in java 6:

private static final long serialversionuid = -6849794470754667710l;

from arraylist class in java 6:

private static final long serialversionuid = 8683452581122892189l;

etc. eclipse offers alternative generate these random numbers (though primary default appears to generate serialversionuid of 1l)

why utilize random numbers? doesn't create more sense start @ 1l , increment 2l when changes sensible revision control? time can think of utilize seemingly random number if didn't specify serialversionuid begin , want (which ties in run-time autogenerated version provide backwards compatibility support).

those "random" numbers numbers would have been generated class automatically in "current" form per java object serialization specification... "current" "current @ time serialversionuid first declared".

that allow info had been serialized still deserialized - while moving forwards more explicit declaration of breaking changes in future.

java serialization