Re: LF to CRLF translation responsibility

2020-05-10 Thread Aaron Schrab
At 11:10 -0400 10 May 2020, Remco Rijnders wrote: Well, the problem might originate in msmtp, but I think opensmtpd also has an issue here in that it breaks / locks up / hangs when presented with this unexpected input, with possible security implications. I'd disagree. Rather than being locked

Re: LF to CRLF translation responsibility

2020-05-10 Thread Remco Rijnders
On Wed, May 06, 2020 at 01:38:35PM -0700, Kevin wrote in <20200506203835.ga30...@afu.lan>: On Wed, May 06, 2020 at 04:12:27PM -0400, Kurt Hackenberg wrote: From reading the ticket, I suspect the problem is in msmtp, an SMTP client, a replacement for one function of sendmail. Sounds like msmtp do

Re: LF to CRLF translation responsibility

2020-05-06 Thread Kevin J. McCarthy
On Wed, May 06, 2020 at 11:23:32PM +0200, ilf wrote: Kurt Hackenberg: I guess msmtp is obsolete for Mutt users anyway, since Mutt now contains an SMTP client. No, msmtp is not obsolete for Mutt users. People have different requirements, and it's good Mutt allows choice. (Mutt is an MUA, not

Re: LF to CRLF translation responsibility

2020-05-06 Thread ilf
Kurt Hackenberg: I guess msmtp is obsolete for Mutt users anyway, since Mutt now contains an SMTP client. No, msmtp is not obsolete for Mutt users. People have different requirements, and it's good Mutt allows choice. (Mutt is an MUA, not an MSA afterall. :) -- sent with msmtp -- ilf If y

Re: LF to CRLF translation responsibility

2020-05-06 Thread Kevin J. McCarthy
On Wed, May 06, 2020 at 04:12:27PM -0400, Kurt Hackenberg wrote: Agreed. I don't know of any expicit documentation of this either, but it seems to be standard practice to convert in both directions between the network form and the local form at the boundary between the network and the local s

Re: LF to CRLF translation responsibility

2020-05-06 Thread Kurt Hackenberg
On 2020-05-06 15:32, Claus Assmann wrote: On Wed, May 06, 2020, Kevin J. McCarthy wrote: I suspect this is the contention point for opensmtpd. When invoking $sendmail directly the headers and delimeter are all just LF. Is this expectation that the MSP perform conversion documented anywhere?

Re: LF to CRLF translation responsibility

2020-05-06 Thread Claus Assmann
On Wed, May 06, 2020, Kevin J. McCarthy wrote: > I suspect this is the contention point for opensmtpd. When invoking > $sendmail directly the headers and delimeter are all just LF. Is this > expectation that the MSP perform conversion documented anywhere? Not in an RFC as that's just a local pr

Re: LF to CRLF translation responsibility

2020-05-06 Thread Kevin J. McCarthy
On Wed, May 06, 2020 at 08:44:29PM +0200, Claus Assmann wrote: On Wed, May 06, 2020, Kevin J. McCarthy wrote: RFC5322 states that headers and the header/body delimiter are supposed to be CRLF, but I (and ticket #235) note that mutt is writing them with just LF. Thank you for your reply, Clau

Re: LF to CRLF translation responsibility

2020-05-06 Thread Claus Assmann
On Wed, May 06, 2020, Kevin J. McCarthy wrote: > RFC5322 states that headers and the header/body delimiter are supposed to be > CRLF, but I (and ticket #235) note that mutt is writing them with just LF. That's for SMTP -- if mutt is talking to an MTA (or MSA) it must use CRLF (though some MSAs fi

LF to CRLF translation responsibility

2020-05-06 Thread Kevin J. McCarthy
I'd appreciate feedback on an area I'm not too clear about. RFC5322 states that headers and the header/body delimiter are supposed to be CRLF, but I (and ticket #235) note that mutt is writing them with just LF. The ticket reports this is causing opensmtpd to choke. However, given that Mutt