No, I didn't test it, because I don't have postfix installed anywhere here. But even testing it doesn't mean my experiment nailed down the actual source code logic or the background for it, hence the question.
The absence of consensus in RFCs to date doesn't prevent the publication of a "best current practice", which is what I was getting at if one could collect data from several popular open source and commercial MTAs and find a common decision about this or other issues. -MSK From: [email protected] [mailto:[email protected]] On Behalf Of Jeroen Geilman Sent: Tuesday, September 28, 2010 3:31 PM To: [email protected] Subject: Re: What's done with malformed headers On 09/28/2010 10:45 PM, Murray S. Kucherawy wrote: (If this would be better asked on postfix-devel, let me know and I'll repost it there.) What does Postfix do with a message that has a malformed header? If for example there's a header that's found to contain a line whose first character is alphanumeric (thus not a continuation) but contains no colon, is this treated as a header field, or does this cause the header to end and the body to begin, or something else? Did you test it ? It's trivial to test with telnet. Postfix silently starts the body of the mail at the first non-header. I'm pretty sure sendmail, for example, ends the header there and begins the body such that the malformed line is the first line of the body, Postfix emulates sendmail behaviour to an uncanny degree, then. If there's consensus among popular MTAs about what to do, maybe this should be documented someplace more formally. The relevant RFCs contain useful information, the most telling of which is that the debate on what to do with malformed mail has not abated since the inception of SMTP in the late 70s. This means that there is no "consensus" - there are too many differing opinions. Postfix obviously falls squarely on the side of "don't drop malformed mail unless it is technically undeliverable", which I find not unreasonable. -- J.
