Re: DKIM fails depending on Content-Transfer-Encoding
Hi Kevin, On 2023-09-07 20:07, Kevin J. McCarthy wrote: > On Thu, Sep 07, 2023 at 11:57:09AM +0200, f...@igh.de wrote: > > Thank you for this hint. Unfortunately it does not work for me. Mutt > > insists in 8bit, maybe depending on my locales. Seems I have to study > > the source code... > > See if 'unset allow_8bit' helps. YES! That solved my problem - thanks a lot and best regards Torsten > > -- > Kevin J. McCarthy > GPG Fingerprint: 8975 A9B3 3AA3 7910 385C 5308 ADEF 7684 8031 6BDA -- Torsten Finke f...@igh.de
Re: DKIM fails depending on Content-Transfer-Encoding
On Thu, Sep 07, 2023 at 11:57:09AM +0200, f...@igh.de wrote: Thank you for this hint. Unfortunately it does not work for me. Mutt insists in 8bit, maybe depending on my locales. Seems I have to study the source code... See if 'unset allow_8bit' helps. -- Kevin J. McCarthy GPG Fingerprint: 8975 A9B3 3AA3 7910 385C 5308 ADEF 7684 8031 6BDA signature.asc Description: PGP signature
Re: DKIM fails depending on Content-Transfer-Encoding
Hello Ed, On 2023-09-06 14:46, Ed Blackman wrote: > On Wed, Sep 06, 2023 at 01:33:30PM +0200, f...@igh.de wrote: > > Can I force Mutt to use quoted-printable or base64 by default for > > encoding of plain text? > > It would be better to fix the DKIM problem, but as I have no idea > about that, I'll tell you how I force q-p encoding: include a > non-ASCII space in my signature. > The space between my first and last name in my signature is a > non-breaking space, eg in HTML. In vim I used > to insert it. The non-ASCII character in the > message body tickles mutt to send the message encoded as > quoted-printable. Thank you for this hint. Unfortunately it does not work for me. Mutt insists in 8bit, maybe depending on my locales. Seems I have to study the source code... Best Regards Torsten > Ed Blackman -- Torsten Finke f...@igh.de
Re: DKIM fails depending on Content-Transfer-Encoding
Hello Raf, On 2023-09-07 10:31, raf via Mutt-users wrote: > On Wed, Sep 06, 2023 at 01:33:30PM +0200, f...@igh.de wrote: > > > Dear Mutt Users > > > > recently I experienced DKIM fails that depend on the > > Content-Transfer-Encoding of messages text part. > > > > Being a german I use to write my messages in german with UTF-8 > > encoding. I prefer plain text. My e-mails are DKIM signed. I have > > checked DKIM to be set up correctly twice. > > > > By default Mutt does 8bit encoding for text/plain. Now I found that > > several (most) of the recipient systems fail to check DKIM. ... > Hi, This has come up recently in the Postfix mailing list. > MTAs can convert 8bit messages when sending to another MTA > that doesn't advertise that it can accept 8bit. If the DKIM > signing happens before the conversion, then subsequent DKIM > checks will fail. Work is being done in Postfix to address > this. I don't know about other MTAs. It seems unlikely that > there are any MTAs that can't accept 8bit messages, but perhaps > there are some that are misconfigured and don't advertise the > fact to other MTAs. that's a good hint. I also suspected some MTA. So I should do further investigation on them. Best Regards Torsten -- Torsten Finke f...@igh.de
Re: DKIM fails depending on Content-Transfer-Encoding
On 2023-09-07, raf via Mutt-users wrote: > On Wed, Sep 06, 2023 at 01:33:30PM +0200, f...@igh.de wrote: > >> Dear Mutt Users >> >> recently I experienced DKIM fails that depend on the >> Content-Transfer-Encoding of messages text part. >> >> Being a german I use to write my messages in german with UTF-8 >> encoding. I prefer plain text. My e-mails are DKIM signed. I have >> checked DKIM to be set up correctly twice. >> >> By default Mutt does 8bit encoding for text/plain. Now I found that >> several (most) of the recipient systems fail to check DKIM. >> >> If I force Mutt to change the encoding from 8bit to 7bit, base64, or >> quoted-printable (using ^E in the compose menu), the DKIM checks >> succeed. >> >> Can I force Mutt to use quoted-printable or base64 by default for >> encoding of plain text? >> >> Does anyone have similar experiences? Is there an explanation for this? >> May there be any interference with the MTA? >> >> Interestingly DKIM checks do not fail if I use non-ASCII characters in >> the subject. Also attachements do not cause DKIM to fail. >> >> Best Regards >> >> T. Finke >> >> -- >> >> T. Finke >> f...@igh.de >> > > Hi, This has come up recently in the Postfix mailing list. > MTAs can convert 8bit messages when sending to another MTA > that doesn't advertise that it can accept 8bit. If the DKIM > signing happens before the conversion, then subsequent DKIM > checks will fail. Work is being done in Postfix to address > this. I don't know about other MTAs. It seems unlikely that > there are any MTAs that can't accept 8bit messages, but perhaps > there are some that are misconfigured and don't advertise the > fact to other MTAs. Has AOL/Yahoo/Verizon/...'s server software been finally fixed from its eternal dance between two different failure modes? (Either replacing non-ascii with ? or messing up the encoding); I think it also misadvertised 8-bit support to MUAs... But maybe that really just affects client connections and does not damage messages received from other servers? -- Nuno Silva
Re: DKIM fails depending on Content-Transfer-Encoding
On Thu, Sep 07, 2023 at 10:31:53AM +1000, raf via Mutt-users wrote: > Hi, This has come up recently in the Postfix mailing list. MTAs can > convert 8bit messages when sending to another MTA that doesn't > advertise that it can accept 8bit. If the DKIM signing happens > before the conversion, then subsequent DKIM checks will fail. Work > is being done in Postfix to address this. I don't know about other > MTAs. It seems unlikely that there are any MTAs that can't accept > 8bit messages, but perhaps there are some that are misconfigured and > don't advertise the fact to other MTAs. My own MTA is configured (not *mis*configured) like that, because I prefer SMTP as it was originally specified :) But I don't block incoming messages based on DKIM failure. -- Ian
Re: DKIM fails depending on Content-Transfer-Encoding
On Wed, Sep 06, 2023 at 04:28:46PM +0200, f...@igh.de wrote: > works perfectly if the subject contains any non-ASCII characters. I think this case is not relevant, because any non-ascii characters in the Subject are converted to use the RFC-2047 scheme, I guess already in mutt. -- Ian
Re: DKIM fails depending on Content-Transfer-Encoding
On Wed, Sep 06, 2023 at 01:33:30PM +0200, f...@igh.de wrote: > Dear Mutt Users > > recently I experienced DKIM fails that depend on the > Content-Transfer-Encoding of messages text part. > > Being a german I use to write my messages in german with UTF-8 > encoding. I prefer plain text. My e-mails are DKIM signed. I have > checked DKIM to be set up correctly twice. > > By default Mutt does 8bit encoding for text/plain. Now I found that > several (most) of the recipient systems fail to check DKIM. > > If I force Mutt to change the encoding from 8bit to 7bit, base64, or > quoted-printable (using ^E in the compose menu), the DKIM checks > succeed. > > Can I force Mutt to use quoted-printable or base64 by default for > encoding of plain text? > > Does anyone have similar experiences? Is there an explanation for this? > May there be any interference with the MTA? > > Interestingly DKIM checks do not fail if I use non-ASCII characters in > the subject. Also attachements do not cause DKIM to fail. > > Best Regards > > T. Finke > > -- > > T. Finke > f...@igh.de > Hi, This has come up recently in the Postfix mailing list. MTAs can convert 8bit messages when sending to another MTA that doesn't advertise that it can accept 8bit. If the DKIM signing happens before the conversion, then subsequent DKIM checks will fail. Work is being done in Postfix to address this. I don't know about other MTAs. It seems unlikely that there are any MTAs that can't accept 8bit messages, but perhaps there are some that are misconfigured and don't advertise the fact to other MTAs. cheers, raf
Re: DKIM fails depending on Content-Transfer-Encoding
On Wed, Sep 06, 2023 at 01:33:30PM +0200, f...@igh.de wrote: > Can I force Mutt to use quoted-printable or base64 by default for > encoding of plain text? It would be better to fix the DKIM problem, but as I have no idea about that, I'll tell you how I force q-p encoding: include a non-ASCII space in my signature. The space between my first and last name in my signature is a non-breaking space, eg in HTML. In vim I used to insert it. The non-ASCII character in the message body tickles mutt to send the message encoded as quoted-printable. -- Ed Blackman
Re: DKIM fails depending on Content-Transfer-Encoding
Dear Jan, On 2023-09-06 15:25, Jan Eden via Mutt-users wrote: > On 2023-09-06 13:33, f...@igh.de wrote: > > > Dear Mutt Users > > > > recently I experienced DKIM fails that depend on the > > Content-Transfer-Encoding of messages text part. > > > > Being a german I use to write my messages in german with UTF-8 > > encoding. I prefer plain text. My e-mails are DKIM signed. I have > > checked DKIM to be set up correctly twice. > > > > By default Mutt does 8bit encoding for text/plain. Now I found that > > several (most) of the recipient systems fail to check DKIM. > > > > If I force Mutt to change the encoding from 8bit to 7bit, base64, or > > quoted-printable (using ^E in the compose menu), the DKIM checks > > succeed. > > > > Can I force Mutt to use quoted-printable or base64 by default for > > encoding of plain text? > > In my experience, mutt chooses the simplest encoding required for each > message, and it is usually better not to intervene. that has been my opinion too. But after several tests I found, that DKIM fails dependent on that very encoding. So now I am convinced, that Mutt's decision is not the best and I am looking for a configuration option to fix that. > > Does anyone have similar experiences? Is there an explanation for this? > > May there be any interference with the MTA? > > At what point do you apply the DKIM signature? If the DKIM signature is > calculated for the complete message (as created by mutt), there should > be no mismatch/DKIM failure. Mutt sends the messages to my MTA and as usual the DKIM signature is then created by that MTA via Milter (OpenDKIM). And that works as a charm as long as the plain text is plain ASCII (no UTF-8). It even works perfectly if the subject contains any non-ASCII characters. Also names of attachements may contain those characters. As soon as there is only one non-ASCII character in the plain text message, Mutt choses 8bit and DKIM fails. BTW.: I have fiddled with evolution - it seems that there exists a similar problem. But I cannot force it to chose another encoding. Again: mutt sucks less :-) Best Regards Torsten -- Torsten Finke f...@igh.de
Re: DKIM fails depending on Content-Transfer-Encoding
On 2023-09-06 13:33, f...@igh.de wrote: > Dear Mutt Users > > recently I experienced DKIM fails that depend on the > Content-Transfer-Encoding of messages text part. > > Being a german I use to write my messages in german with UTF-8 > encoding. I prefer plain text. My e-mails are DKIM signed. I have > checked DKIM to be set up correctly twice. > > By default Mutt does 8bit encoding for text/plain. Now I found that > several (most) of the recipient systems fail to check DKIM. > > If I force Mutt to change the encoding from 8bit to 7bit, base64, or > quoted-printable (using ^E in the compose menu), the DKIM checks > succeed. > > Can I force Mutt to use quoted-printable or base64 by default for > encoding of plain text? In my experience, mutt chooses the simplest encoding required for each message, and it is usually better not to intervene. > Does anyone have similar experiences? Is there an explanation for this? > May there be any interference with the MTA? At what point do you apply the DKIM signature? If the DKIM signature is calculated for the complete message (as created by mutt), there should be no mismatch/DKIM failure. - Jan signature.asc Description: PGP signature