> > Under no circumstances it should ever happen that "high secret > > attachment" of user A has the same checksum of "funny pic" > from user > > B, and the DB makes a link for user B to "high secret > attachment" of > > user A. > > Definitely. We need an algorithm that will prevent collisions > at all cost. Such checksums will only be calculated during > insertion to ensure uniqueness of the attachment so they will > not affect the retrieval processes (pop/imap). > > That said, "high secret" stuff should of course have been > encrypted to begin with.
How about just taking x size hunk from start and from end of the file, then md5sum both of them? I think that possibility that you will get 2 exactly same md5 hash is quite zero.. Bonus comes that you don't need to do md5 hash from whole file, so server should not suffer performance loss because of this.
BEGIN:VCARD VERSION:2.1 N:Partanen;Jani FN:[EMAIL PROTECTED] ORG:Pelikaista / Mediatraffic Oy TITLE:Battlefield admin TEL;CELL;VOICE:+358 505 817 817 ADR;WORK:;;Kirjolohentie 8 a2;Saimaanharju;;54915;Finland LABEL;WORK;ENCODING=QUOTED-PRINTABLE:Kirjolohentie 8 a2=0D=0ASaimaanharju 54915=0D=0AFinland ADR;HOME:;;Kirjolohentie 8 a2;Saimaanharju;;54915;Finland LABEL;HOME;ENCODING=QUOTED-PRINTABLE:Kirjolohentie 8 a2=0D=0ASaimaanharju 54915=0D=0AFinland URL;WORK:http://www.pelikaista.net EMAIL;PREF;INTERNET:[EMAIL PROTECTED] REV:20060812T221011Z END:VCARD
_______________________________________________ DBmail mailing list [email protected] https://mailman.fastxs.nl/mailman/listinfo/dbmail
