Loading ...
Sorry, an error occurred while loading the content.

Re: valid_hostname chokes on trailing dot

Expand Messages
  • Chad Whitacre
    ... Thanks for the pointers. ... Right, sorry ... behavior. :^P chad
    Message 1 of 20 , Oct 1, 2008
    • 0 Attachment
      > I suggest bringing this on an IETF mailing list.

      Thanks for the pointers.


      > calling it a bug without confirmation by the developpers is
      > sure to help your cause ;-p

      Right, sorry ... "behavior." :^P




      chad
    • Chad Whitacre
      Noel, ... Thanks for the welcome and info. I ve updated the Wikipedia entry on FQDN s accordingly: http://en.wikipedia.org/wiki/FQDN chad
      Message 2 of 20 , Oct 1, 2008
      • 0 Attachment
        Noel,

        > The current behavior is correct and won't be changed.

        Thanks for the welcome and info. I've updated the Wikipedia entry
        on FQDN's accordingly:

        http://en.wikipedia.org/wiki/FQDN


        chad
      • Wietse Venema
        ... Please cite the relevant section of the relevant RFC. Wietse
        Message 3 of 20 , Oct 1, 2008
        • 0 Attachment
          Chad Whitacre:
          > Greetings,
          >
          > In src/util/valid_hostname.c (version 2.5.5), valid_hostname
          > gives a "misplaced delimiter" warning if the hostname ends in a
          > trailing dot. Since a trailing dot is technically part of a FQDN,
          > I suggest that valid_hostname allow them. Patch attached.

          Please cite the relevant section of the relevant RFC.

          Wietse
        • Chad Whitacre
          Happy to if you point me to it. I m not an expert. chad
          Message 4 of 20 , Oct 1, 2008
          • 0 Attachment
            Happy to if you point me to it. I'm not an expert.


            chad


            Wietse Venema wrote:
            > Chad Whitacre:
            >> Greetings,
            >>
            >> In src/util/valid_hostname.c (version 2.5.5), valid_hostname
            >> gives a "misplaced delimiter" warning if the hostname ends in a
            >> trailing dot. Since a trailing dot is technically part of a FQDN,
            >> I suggest that valid_hostname allow them. Patch attached.
            >
            > Please cite the relevant section of the relevant RFC.
            >
            > Wietse
            >
          • Chad Whitacre
            ... Is this the right place? https://tools.ietf.org/html/rfc5321#section-2.3.5 chad
            Message 5 of 20 , Oct 1, 2008
            • 0 Attachment
              >> Please cite the relevant section of the relevant RFC.
              > Happy to if you point me to it. I'm not an expert.

              Is this the right place?

              https://tools.ietf.org/html/rfc5321#section-2.3.5



              chad
            • John Peach
              On Wed, 01 Oct 2008 12:40:57 -0400 ... If it is, it does not back up your assertion that a trailing dot is part of the FQDN. -- John
              Message 6 of 20 , Oct 1, 2008
              • 0 Attachment
                On Wed, 01 Oct 2008 12:40:57 -0400
                Chad Whitacre <chad@...> wrote:

                > >> Please cite the relevant section of the relevant RFC.
                > > Happy to if you point me to it. I'm not an expert.
                >
                > Is this the right place?
                >
                > https://tools.ietf.org/html/rfc5321#section-2.3.5
                >

                If it is, it does not back up your assertion that a trailing dot is
                part of the FQDN.


                --
                John
              • Wietse Venema
                ... Sorry, that is backwards. If you want to correct Postfix behavior, then you need to provide supporting documentation that Postfix is in error. So please
                Message 7 of 20 , Oct 1, 2008
                • 0 Attachment
                  Chad Whitacre:
                  > Greetings,
                  >
                  > In src/util/valid_hostname.c (version 2.5.5), valid_hostname
                  > gives a "misplaced delimiter" warning if the hostname ends in a
                  > trailing dot. Since a trailing dot is technically part of a FQDN,
                  > I suggest that valid_hostname allow them. Patch attached.

                  Wietse:
                  > > Please cite the relevant section of the relevant RFC.

                  Chad Whitacre:
                  > Happy to if you point me to it. I'm not an expert.

                  Sorry, that is backwards.

                  If you want to "correct" Postfix behavior, then you need to provide
                  supporting documentation that Postfix is in error.

                  So please point out the relevant sections of email related RFCs,
                  otherwise I suggest that you refrain from posting "corrections".

                  Wietse
                • Chad Whitacre
                  John, ... Right, I stopped asserting that when Wietse and Noel told me that FQDN means something different for SMTP than for DNS. The question (I thought) was
                  Message 8 of 20 , Oct 1, 2008
                  • 0 Attachment
                    John,

                    > If it is, it does not back up your assertion that a trailing
                    > dot is part of the FQDN.

                    Right, I stopped asserting that when Wietse and Noel told me that
                    FQDN means something different for SMTP than for DNS. The
                    question (I thought) was what to point to from Wikipedia.


                    chad



                    John Peach wrote:
                    >
                    > On Wed, 01 Oct 2008 12:40:57 -0400
                    > Chad Whitacre <chad@...> wrote:
                    >
                    >> >> Please cite the relevant section of the relevant RFC.
                    >> > Happy to if you point me to it. I'm not an expert.
                    >>
                    >> Is this the right place?
                    >>
                    >> https://tools.ietf.org/html/rfc5321#section-2.3.5
                    >>
                    >
                    > If it is, it does not back up your assertion that a trailing dot is
                    > part of the FQDN.
                    >
                    >
                  • Wietse Venema
                    ... Does the text say SEPARATED by dots or TERMINATED by dot ? Wietse
                    Message 9 of 20 , Oct 1, 2008
                    • 0 Attachment
                      Chad Whitacre:
                      > Is this the right place?
                      >
                      > https://tools.ietf.org/html/rfc5321#section-2.3.5

                      Does the text say "SEPARATED by dots" or "TERMINATED by dot"?

                      Wietse
                    • Chad Whitacre
                      Wietse, ... We ve got a misunderstanding. I accept your statement that FQDN means something different in SMTP than DNS. I m not trying to correct Postfix
                      Message 10 of 20 , Oct 1, 2008
                      • 0 Attachment
                        Wietse,

                        > Sorry, that is backwards.
                        >
                        > If you want to "correct" Postfix behavior, then you need to
                        > provide supporting documentation that Postfix is in error.
                        >
                        > So please point out the relevant sections of email related
                        > RFCs, otherwise I suggest that you refrain from posting
                        > "corrections".

                        We've got a misunderstanding. I accept your statement that FQDN
                        means something different in SMTP than DNS. I'm not trying to
                        correct Postfix behavior, I'm trying to find the best reference
                        to add to Wikipedia's article on FQDNs. If I'm not mistaken, 5321
                        2.3.5 is this reference (and, coincidentally, it appears that
                        5321 is being published as we speak). Yes?



                        chad
                      • Noel Jones
                        ... Yes, that looks like an appropriate link. it s certainly a better reference than this wandering thread... -- Noel Jones
                        Message 11 of 20 , Oct 1, 2008
                        • 0 Attachment
                          Chad Whitacre wrote:
                          > Wietse,
                          >
                          > > Sorry, that is backwards.
                          > >
                          > > If you want to "correct" Postfix behavior, then you need to
                          > > provide supporting documentation that Postfix is in error.
                          > >
                          > > So please point out the relevant sections of email related
                          > > RFCs, otherwise I suggest that you refrain from posting
                          > > "corrections".
                          >
                          > We've got a misunderstanding. I accept your statement that FQDN means
                          > something different in SMTP than DNS. I'm not trying to correct Postfix
                          > behavior, I'm trying to find the best reference to add to Wikipedia's
                          > article on FQDNs. If I'm not mistaken, 5321 2.3.5 is this reference
                          > (and, coincidentally, it appears that 5321 is being published as we
                          > speak). Yes?
                          >
                          >
                          >
                          > chad


                          Yes, that looks like an appropriate link.
                          it's certainly a better reference than this wandering thread...

                          --
                          Noel Jones
                        • Wietse Venema
                          ... A mis-understanding as in posting a patch that corrects Postfix behavior without adequate justification. According to RFC 5312 section 2.3.5, SMTP uses
                          Message 12 of 20 , Oct 1, 2008
                          • 0 Attachment
                            Chad Whitacre:
                            > Wietse,
                            >
                            > > Sorry, that is backwards.
                            > >
                            > > If you want to "correct" Postfix behavior, then you need to
                            > > provide supporting documentation that Postfix is in error.
                            > >
                            > > So please point out the relevant sections of email related
                            > > RFCs, otherwise I suggest that you refrain from posting
                            > > "corrections".
                            >
                            > We've got a misunderstanding. I accept your statement that FQDN
                            > means something different in SMTP than DNS. I'm not trying to
                            > correct Postfix behavior, I'm trying to find the best reference
                            > to add to Wikipedia's article on FQDNs. If I'm not mistaken, 5321
                            > 2.3.5 is this reference (and, coincidentally, it appears that
                            > 5321 is being published as we speak). Yes?

                            A mis-understanding as in posting a patch that "corrects" Postfix
                            behavior without adequate justification.

                            According to RFC 5312 section 2.3.5, SMTP uses the dot as SEPARATOR
                            not TERMINATOR in domain names.

                            Wietse
                          • mouss
                            ... Does DNS even define a hostname ? My understanding is that this is a natural term, not a technically defined one (after all, on many systems, the
                            Message 13 of 20 , Oct 1, 2008
                            • 0 Attachment
                              Noel Jones wrote:
                              > Welcome to postfix!
                              >
                              > Valid hostnames in mail are not the same as valid DNS hostnames. Mail
                              > hostnames never end with a dot.

                              Does DNS even define a "hostname"? My understanding is that this is a
                              "natural" term, not a technically defined one (after all, on many
                              systems, the "hostname" is not FQDN, et names like "foo_bar" are ok,
                              eventhough they can't be used as "mail" hostnames).

                              >
                              > The current behavior is correct and won't be changed.
                              >
                              > but thanks for your willingness to help and taking the effort to supply
                              > a patch.
                              >
                            • Chad Whitacre
                              Wietse, ... o I raised a legitimate issue. SMTP s use of FQDN is not widely known. Afaict, it was not even formally defined until RFC 5321 was published
                              Message 14 of 20 , Oct 1, 2008
                              • 0 Attachment
                                Wietse,

                                > A mis-understanding as in posting a patch that "corrects"
                                > Postfix behavior without adequate justification.
                                >
                                > According to RFC 5312 section 2.3.5, SMTP uses the dot as
                                > SEPARATOR not TERMINATOR in domain names.

                                o I raised a legitimate issue. SMTP's use of "FQDN" is not
                                widely known. Afaict, it was not even formally defined until
                                RFC 5321 was published today--during our conversation.

                                o I provided a patch.

                                o I apologized quickly for calling this behavior a bug.

                                o I never used the word "corrects" or otherwise implied that
                                Postfix has it wrong.

                                o I accepted your and Noel's clarifications about the use of
                                "FQDN" in SMTP.

                                o I followed your advice in searching the IETF-SMTP list for
                                more info.

                                o I updated the Wikipedia article on FQDN's accordingly.

                                o Here was my misunderstanding: I thought you posted the
                                message about RFCs 5321 and 5322, and were asking me to
                                cite these in the Wikipedia article. My fault, sorry.

                                o I asked for help in finding the specific reference.

                                o I found the reference myself.

                                o I updated Wikipedia again.


                                I'm human, I make mistakes, and I hope I apologize for them. But
                                I'm not an idiot and I'm not malicious. I'm on your side. Don't
                                you think you could lighten up a little?



                                chad




                                Wietse Venema wrote:
                                > Chad Whitacre:
                                >> Wietse,
                                >>
                                >> > Sorry, that is backwards.
                                >> >
                                >> > If you want to "correct" Postfix behavior, then you need to
                                >> > provide supporting documentation that Postfix is in error.
                                >> >
                                >> > So please point out the relevant sections of email related
                                >> > RFCs, otherwise I suggest that you refrain from posting
                                >> > "corrections".
                                >>
                                >> We've got a misunderstanding. I accept your statement that FQDN
                                >> means something different in SMTP than DNS. I'm not trying to
                                >> correct Postfix behavior, I'm trying to find the best reference
                                >> to add to Wikipedia's article on FQDNs. If I'm not mistaken, 5321
                                >> 2.3.5 is this reference (and, coincidentally, it appears that
                                >> 5321 is being published as we speak). Yes?
                                >
                                > A mis-understanding as in posting a patch that "corrects" Postfix
                                > behavior without adequate justification.
                                >
                                > According to RFC 5312 section 2.3.5, SMTP uses the dot as SEPARATOR
                                > not TERMINATOR in domain names.
                                >
                                > Wietse
                                >
                              • Noel Jones
                                ... An interesting observation. Yes, I believe the dozens of DNS related RFCs mostly talk about domain name (aka. FQDN) and host address (referring to
                                Message 15 of 20 , Oct 1, 2008
                                • 0 Attachment
                                  mouss wrote:
                                  > Noel Jones wrote:
                                  >> Welcome to postfix!
                                  >>
                                  >> Valid hostnames in mail are not the same as valid DNS hostnames.
                                  >> Mail hostnames never end with a dot.
                                  >
                                  > Does DNS even define a "hostname"? My understanding is that this is a
                                  > "natural" term, not a technically defined one (after all, on many
                                  > systems, the "hostname" is not FQDN, et names like "foo_bar" are ok,
                                  > eventhough they can't be used as "mail" hostnames).

                                  An interesting observation. Yes, I believe the dozens of DNS
                                  related RFCs mostly talk about "domain name" (aka. FQDN) and
                                  "host address" (referring to the IP) rather than "hostname",
                                  but this is getting sufficiently OT that I don't really care.

                                  --
                                  Noel Jones
                                • mouss
                                  ... Actually 5321 didn t change the definition of helo (and even 2821 didn t really change it). ... that s nice and constructive. ... sure, but it can t be
                                  Message 16 of 20 , Oct 1, 2008
                                  • 0 Attachment
                                    Chad Whitacre wrote:
                                    > Wietse,
                                    >
                                    > > A mis-understanding as in posting a patch that "corrects"
                                    > > Postfix behavior without adequate justification.
                                    > >
                                    > > According to RFC 5312 section 2.3.5, SMTP uses the dot as
                                    > > SEPARATOR not TERMINATOR in domain names.
                                    >
                                    > o I raised a legitimate issue. SMTP's use of "FQDN" is not
                                    > widely known. Afaict, it was not even formally defined until
                                    > RFC 5321 was published today--during our conversation.
                                    >

                                    Actually 5321 didn't change the definition of helo (and even 2821 didn't
                                    really change it).

                                    > o I provided a patch.

                                    that's nice and constructive.

                                    >
                                    > o I apologized quickly for calling this behavior a bug.

                                    sure, but it can't be removed from the history file ;-p

                                    >
                                    > o I never used the word "corrects" or otherwise implied that
                                    > Postfix has it wrong.
                                    >
                                    > o I accepted your and Noel's clarifications about the use of
                                    > "FQDN" in SMTP.
                                    >
                                    > o I followed your advice in searching the IETF-SMTP list for
                                    > more info.

                                    do you mix mail from me and mail from Wietse?

                                    >
                                    > o I updated the Wikipedia article on FQDN's accordingly.
                                    >
                                    > o Here was my misunderstanding: I thought you posted the
                                    > message about RFCs 5321 and 5322, and were asking me to
                                    > cite these in the Wikipedia article. My fault, sorry.
                                    >
                                    > o I asked for help in finding the specific reference.
                                    >
                                    > o I found the reference myself.
                                    >
                                    > o I updated Wikipedia again.
                                    >
                                    >
                                    > I'm human, I make mistakes, and I hope I apologize for them. But I'm not
                                    > an idiot and I'm not malicious. I'm on your side. Don't you think you
                                    > could lighten up a little?

                                    don't be upset by the "style". the trailing dot thing has already been
                                    debated here a long time ago. so bringing it again does annoy those who
                                    have seen the old thread. While you're not required to know all the
                                    posts and discussions that happened here, don't be surprised if the
                                    responses you get aren't as "smooth" as you would like. but there's no
                                    hate or animosity.

                                    Anyway, as Noel said, "thanks for your willingness to help...".
                                  • Wietse Venema
                                    ... I am enlighted, but this does not always come across in my short responses. Wietse
                                    Message 17 of 20 , Oct 1, 2008
                                    • 0 Attachment
                                      Chad Whitacre:
                                      > Wietse,
                                      >
                                      > > A mis-understanding as in posting a patch that "corrects"
                                      > > Postfix behavior without adequate justification.
                                      > >
                                      > > According to RFC 5312 section 2.3.5, SMTP uses the dot as
                                      > > SEPARATOR not TERMINATOR in domain names.
                                      >
                                      > o I raised a legitimate issue. SMTP's use of "FQDN" is not
                                      > widely known. Afaict, it was not even formally defined until
                                      > RFC 5321 was published today--during our conversation.
                                      >
                                      > o I provided a patch.
                                      >
                                      > o I apologized quickly for calling this behavior a bug.
                                      >
                                      > o I never used the word "corrects" or otherwise implied that
                                      > Postfix has it wrong.
                                      >
                                      > o I accepted your and Noel's clarifications about the use of
                                      > "FQDN" in SMTP.
                                      >
                                      > o I followed your advice in searching the IETF-SMTP list for
                                      > more info.
                                      >
                                      > o I updated the Wikipedia article on FQDN's accordingly.
                                      >
                                      > o Here was my misunderstanding: I thought you posted the
                                      > message about RFCs 5321 and 5322, and were asking me to
                                      > cite these in the Wikipedia article. My fault, sorry.
                                      >
                                      > o I asked for help in finding the specific reference.
                                      >
                                      > o I found the reference myself.
                                      >
                                      > o I updated Wikipedia again.
                                      >
                                      >
                                      > I'm human, I make mistakes, and I hope I apologize for them. But
                                      > I'm not an idiot and I'm not malicious. I'm on your side. Don't
                                      > you think you could lighten up a little?

                                      I am enlighted, but this does not always come across in my
                                      short responses.

                                      Wietse
                                    Your message has been successfully submitted and would be delivered to recipients shortly.