ietf-822
[Top] [All Lists]

Re: Comment on encapsulation BNF in latest RFC-XXXX

1991-10-28 11:13:25
checksum:

Checksums are useful when there is a reasonable basis for fearing loss
of data integrity.  

This devolves to asking whether there is a reasonable basis for
believing that the email internet is likely to corrupt message content.

This seems like a no-brainer, to me, so adding checksum to all/part of
the body content(s) seems like a win.

One could choose to make it optional, as per UDP, rather than required,
as per TCP.  On the whole, however, I view optional checksums as a very
bad idea, per the experience of NFS defaulting to no checksums and
thereby experiencing data loss when used in a wide area net.  (I'm told
that NFS now defaults to using checksum.)

As to the choice of checksum, CRC is probably overkill.  MD4 or MD5 seems
to be getting quite popular.  The Security Area probably can give quick
guidance about the best choice.

Dave

<Prev in Thread] Current Thread [Next in Thread>