On Fri, Oct 18, 2002 at 02:34:50AM +0200, Jeroen van Erp wrote: > On Thu, 17 Oct 2002 23:17:19 +0200 (CEST) > Rudi Sluijtman <rudi@xxxxxxxxxxxxx> wrote: >> waarom maak je van simpele ascii-tekst een base64 gecodeerde attachment? > Dat heb ik inderdaad een keer meer gehoord en ik heb voor de > verandering geen flauw idee. Ik doe niets meer dan mijn emails > signen met mijn gpg key, en gebruik verder Sylpheed als mailer. Here is the idea: Any message containing high-bit characters could be recoded (e.g. to Quoted Printables) by slightly buggy MTAs, if they are eight-bit safe, but they detect the next MTA in the chain is not. (They are not supposed to touch multipart/encrypted or multipart/signed messages at all, but some do) In your message, it was the "é" of "Café" in the text you quoted. So, for extra safety, it is mandatory that clear-signed messages be constrained to 7-bit characters only (section 3 of RFC 3156), if necessary by QP or base64 encoding. Now, Base64 encoding might make sense for Japanese, because nearly every character is high-bit (so, if you use QP and the reader's MUA is not QP-capable, he is screwed anyway, too much information is lost), but IMHO isn't for European languages, where only a fairly low rate of characters is high-bit, and thus QP-encoded mail can be read pretty easily even if the MUA is not QP-capable. I've made patches to address this, included my next message. Apply to you heart's desire.
<<inline: application/pgp-signature>>