Eliot Lear wrote: > I now understand what you mean by outsourced. Sorry if that was unclear, I meant "errata not maintained by the RFC editor". > Most of these comments are truly editorial in nature, but > that happens with FULL standards as well. Agree, but while it's no full standard it would be integrated, the differences between 4409 and 2476 (or 4234 and 2234) are also mainly editorial. For 4422 they stayed at PS, maybe they considered the changes as too substantial. 2396 and 3986 on the other hand were promoted on standards track, although the ABNF was rewritten from scratch twice. Maybe ABNF changes are considered as irrelevant. One obscure <toplablel> killed, good riddance. [2821] > All of those are typos that people can work around without > great difficulty. Yes, but at some point in time I hope we nail this <toplabel> issue. To some degree 3696 already does this, but it's "only" prose, it doesn't offer "updates 1123 2.1 and 1034", and it should be a BCP or on standards track. > There are many instances when an MTA can and should accept > mail when it is not known to be deliverable. The most > obvious example is when it is acting as an MSA. ACK, when it knows that MAIL FROM and originator are related, e.g. an MSA implementing 4409 6.1, it can accept the mail for forward or delivery. > It is certainly the case for nearly every mail I send or > receive. For the latter you're apparently behind a good spam filter and never see the 84% spam. Insert your favourite number, I don't insist on 84%. I also don't insist on 11% misdirected bounces. My personal inbox statistics would be more like 98 +1 +1%, but that's skewed by several factors, almost no mailing lists, no identified worms, most phishes only arriving as "we've deleted something" info, and SPF. Details for the SMTP list, 2821 can't be promoted on standards track as is. We've got a full standard working in theory, but not more used as designed, a PS not working as designed, but used almost everywhere, and an experiment fixing it for those who want it (in essence the least common denominator of the STD and the PS, offering hints when accepting mail from unknown strangers should work as in the "good old times" before spam). > Mailing lists SHOULD modify the mail from Yes, 1123 5.3.6(b) and other 5.3.x are fine, only 5.3.6(a) doesn't work as expected without source routes. Apparently the spammers needed some years to figure this out, it wasn't as bad as today when 2821 was published. But the whole spam problem isn't as old as RFC 1123, when did this plague start, 199x in Usenet ? Frank _______________________________________________ Ietf@xxxxxxxx https://www1.ietf.org/mailman/listinfo/ietf