https://bz.apache.org/bugzilla/show_bug.cgi?id=55808
--- Comment #3 from Tom Fredrik Blenning <bfg-...@blenning.no> --- The notion in this bugreport that MD5 and SHA1 are too weak shows a complete lack of understanding of what the threat scenario really is. If you choose to use these checksums for anything cryptographic you would have to be insane. Upgrading these checksums would be security theater. If anything I would suggest weakening them, as their real value, as suggested by Rüdiger, are to guard against network errors and similar errors. For this purpose the currently used algorithms are overkill, since this could be done by non-cryptographic hashes that are better suited, however these two algorithms are useful since they already are present as shell commands in just about any system you would be likely to use. Should you as suggested by the reporter increase the cryptographic strength of these hashes, there would be no real benefit. A possible attack would be a MitM attack. Let's assume that someone is able to pull this off, since we are on a single channel, the attacker would with no significant extra effort be able to change the hashes, for any cryptograhical strength, there would be no added benefit. On the other hand, with PGP, there are many possibilities for transmitting the key, both in band, but time-shifted, and out-of-band using keyservers, providing security against malicious MitM. If you upgrade them, you risk loosing compatibility again for no real benefit. In my not so humble opinion, I would have closed this issue as invalid, but that's just me. -- You are receiving this mail because: You are the assignee for the bug. --------------------------------------------------------------------- To unsubscribe, e-mail: docs-unsubscr...@httpd.apache.org For additional commands, e-mail: docs-h...@httpd.apache.org
