Dear piler users,
recently someone (disclaimer: he does not use piler) has raised a topic
that the deduplication in piler may cause data loss.
As you may know, piler calculates an sha256 hash value for all
attachments,
and compares them against the attachment table. If the hash is found,
then
piler thinks it's an already stored copy of the same attachment, and
discards
it.
The problem is that all hash functions come with the chance of
collision, ie.
when two different data provide the very same hash value. If it
happened, then
piler would lose the given attachment.
I found an article on stackoverflow trying to estimate the potential
and actual
chance of such an unfortunate accident:
http://stackoverflow.com/questions/4014090/is-it-safe-to-ignore-the-possibility-of-sha-collisions-in-practice
However I'm interested in your opinion whether you sleep well, or you
are
paranoid of this possibility of attachment loss.
Janos
