Cameron Simpson wrote on 07/09/12 at 08:41:26 +1000: > On 08Jul2012 18:32, Jack M <j...@forallx.net> wrote: > | However, I do keep copies of my sent > | mail in $record, and when I look at the saved copy, it is not QP-encoded. > | Also, the mysterious QP only happens on *some* mails that I send, indeed, > only > | on *some* of the mails with non-ASCII in them. This makes me think mutt > isn't > | the culprit. > > It does sound that way. Is it associated with particular target addresses? > That would involve particular mail systems in the transport.
The phenomenon hasn't yet happened enough for me to detect a pattern, although the instance that made me start this thread was a message I sent to a googlegroups mailing list; it was fine when I sent it, but when the list sent me a copy, it was QP-encoded. > Also, for a message that _was_ QP encoded badly, can you show us the exact > headers as they are in your copy in $record? Mutt may not be doing the QP, > but it may be releasing text that is open to wrong QP markup (I really mean > the charset labelling here, not the bytes). It turns out that there was no bogus QP encoding, at least not in the sense of 'bogus' that would mean "they attemtped to QP-encode it but screwed it up in the process". I had earlier reported that the message in question was utf8 that was (unexpectedly) QP-encoded, but reported by the headers as being latin1. I was wrong about this; the headers did say utf8 after all. This whole half of my problem was due to my QP-decoding apparatus in Vim not being able to handle multibyte. > Um, one more thing. UTF-8 includes bytes with the high bit set for some > data, including your example problematic character. Mail systems _are_ > allowed to rewrite content into 7-bit safe forms for transport, and that is > probably what is triggering the QP encoding. Yes, this seems likely to me too. > Can you reliably reproduce the issue? No, not yet anyway. If I eventually can, I'll try your suggestion of forcing mutt to emit utf-7; I presume that's what $send_charset is for. --Jack