|
From: Matthias A. <mat...@gm...> - 2026-08-31 20:32:45
|
Am 31.08.26 um 10:42 schrieb Andrea Venturoli: > On 9/17/25 09:18, Andrea Venturoli wrote: >> On 6/23/25 18:53, Matthias Andree via Fetchmail-users wrote: >> >>> I would probably start by checking what actual error code is passed >>> through this code in driver.c (near line 1150) as return from the - >>> >getauth(...) method: >> >> Hello. >> >> I think I found the reason. >> In the logs I see: >> >>> sm-mta[13685]: 589NKPid013685: collect: relay=[127.0.2.1], >>> from=<FET...@ma...>, info=Bare linefeed (LF) >>> not allowed, where=body, status=tempfail >>> sm-mta[13685]: 589NKPid013685: from=<FETCHMAIL- >>> DA...@ma...>, size=337, class=0, nrcpts=1, >>> proto=ESMTP, daemon=MTA, relay=[127.0.2.1] >> >> >> >> I'll try and change the code to use CRLF instead... > Hello. > Sorry for taking so long. > I fixed this with the two attached patches. Hello Andrea, thanks a lot for the debugging and the patch proposal. You wouldn't need to patch the "report(...)" lines reporting to stderr, but you seem to have a found a long-standing bug that we don't send CR on strict SMTP/LMTP listeners... whereas a native MDA (maildrop, deliver, or similar) or something called via Unix-domain socket might not want CRs but bare LF. We do have a forcecr option in fetchmail, but I think I need to auto-enable it for sending warnings out via SMTP... and therefore I won't take the patches as they are. Filed my reminder at https://gitlab.com/fetchmail/fetchmail/-/work_items/97 - I need to fix this another day. Thanks again. Regards, Matthias |