Hi, I have some trouble with sups behavior in a certain case. Some clients seem to put the mailinglist in the CC field. Apparently if someone with such a client replies to me, sup will automatically reply to the person, not to the list. In cases where such a client replies to another person on the list, sup will reply to the list. Example 1: from: A to: me CC: list reply: from: me to: A Example 2: from: A to: B CC: list reply: from: me to: list I want it to behave like in Example 2 in the case of Example 1 as well. Is there a way to do that? It''s really annoying to send to single people instead of the list by accident. Regards, -- Philipp -- "Wir stehen selbst entt?uscht und sehn betroffen / Den Vorhang zu und alle Fragen offen." Bertolt Brecht, Der gute Mensch von Sezuan
Excerpts from Philipp''s message of Thu Aug 19 06:52:56 -0400 2010:> It''s really annoying to send to single people instead of the list by > accident.It''s really annoying to send a message to the list instead of to a single person by accident. In any case, mailing lists can set a Reply-To header, and I think Sup will reply to the address included in that header. If it does not, then I think that''s a bug :-). Cheers, Israel
Excerpts from Israel Herraiz''s message of Thu Aug 19 07:58:29 -0400 2010:> Excerpts from Philipp''s message of Thu Aug 19 06:52:56 -0400 2010: > > It''s really annoying to send to single people instead of the list by > > accident. > > It''s really annoying to send a message to the list instead of to a > single person by accident. > > In any case, mailing lists can set a Reply-To header, and I think Sup > will reply to the address included in that header. If it does not, > then I think that''s a bug :-). > > Cheers, > IsraelI partially agree. I think we can take a lesson from gmail here: A simple reply always replies to the sender. A group reply will include the list of course, and in that case you are responsible for trimming the recipients. I''ve been caught by surprise when in ''sup a simple reply was sent to the list *instead* of the sender. It is hard to predict when it will happen, because the Reply-To header is not ordinarily visible. Erin Scott Sheldon
Excerpts from Israel Herraiz''s message of 2010-08-19 13:58:29 +0200:> Excerpts from Philipp''s message of Thu Aug 19 06:52:56 -0400 2010: > > It''s really annoying to send to single people instead of the list by > > accident. > > It''s really annoying to send a message to the list instead of to a > single person by accident. > > In any case, mailing lists can set a Reply-To header, and I think Sup > will reply to the address included in that header. If it does not, > then I think that''s a bug :-). > > Cheers, > IsraelI''m not sure that it''s not a header issue. I don''t know a lot about headers, but I think you can''t assume that a reply to is set. It isn''t on the list were I just experienced the problem and it isn''t on this list either. I found ''In-reply-to'' but no ''Reply-to''. No idea which other headers could be relevant. ''Precedence: list'' sounded like it could be relevant, but it''s there in all cases. -- Philipp -- "Wir stehen selbst entt?uscht und sehn betroffen / Den Vorhang zu und alle Fragen offen." Bertolt Brecht, Der gute Mensch von Sezuan
Philipp, 2010-08-19 13:52:> Some clients seem to put the mailinglist in the CC field.Which is pretty annoying.> Apparently if someone with such a client replies to me, sup will > automatically reply to the person, not to the list.This might (and I really am guessing here) be because sup skips duplicates (by message id). In the first exaple you get the mail twice (assuming you are subscribed to lista). The one you most probably get first (and gets indexed first) is the message you get directly and which does not have list headers. The one which makes the extra hop via list server (and has the list headers) gets to you later and is simply discarded as duplicate. When you go reply the mail, sup is perfectly unaware of it being a list mail.> It''s really annoying to send to single people instead of the list by > accident.I can tell you it is _way_ more annoying to accidentally send private mail to list (I''ve done that, once _really_ badly). So this behavior (automatical selection between list reply and private reply) should IMO be conservative in this sense. -- Tero Tilus ## 050 3635 235 ## http://tero.tilus.net/
Excerpts from Erin Sheldon''s message of 2010-08-19 14:27:04 +0200:> Excerpts from Israel Herraiz''s message of Thu Aug 19 07:58:29 -0400 2010: > > Excerpts from Philipp''s message of Thu Aug 19 06:52:56 -0400 2010: > > > It''s really annoying to send to single people instead of the list by > > > accident. > > > > It''s really annoying to send a message to the list instead of to a > > single person by accident. > > > > In any case, mailing lists can set a Reply-To header, and I think Sup > > will reply to the address included in that header. If it does not, > > then I think that''s a bug :-). > > > > Cheers, > > Israel > > I partially agree. > > I think we can take a lesson from gmail here: A simple reply always > replies to the sender. A group reply will include the list of course, > and in that case you are responsible for trimming the recipients. I''ve > been caught by surprise when in ''sup a simple reply was sent to the list > *instead* of the sender. It is hard to predict when it will happen, > because the Reply-To header is not ordinarily visible. > > Erin Scott SheldonI''d like it the other way around, always reply to list and simply change to ''sender'' in the (at least for me) rare cases where it''s not supposed to go to the list. -- Philipp -- "Wir stehen selbst entt?uscht und sehn betroffen / Den Vorhang zu und alle Fragen offen." Bertolt Brecht, Der gute Mensch von Sezuan
Israel Herraiz, 2010-08-19 14:58:> It''s really annoying to send a message to the list instead of to a > single person by accident. > > In any case, mailing lists can set a Reply-To headerWhoa! Mailing lists _can_ do that, but they should not. Ironically you happend to mention one of the reasons why they should not. :D Let''s consider the max damage done by modifying vs. not modifying reply-to: If the header is unmangled the user may in certain sircumstances accidentally send private mail instead of list mail. That''s an annoyance at most. If the header is mangled by list server the user may in certain sircumstances accidentally send list mail instead of private mail. That may cost the user a lot of goodwill, money or her job. Ergo, list setting reply-to header violates the principle of least damage. -- Tero Tilus ## 050 3635 235 ## http://tero.tilus.net/
Excerpts from Tero Tilus''s message of 2010-08-19 14:33:24 +0200:> Philipp, 2010-08-19 13:52: > > Some clients seem to put the mailinglist in the CC field. > > Which is pretty annoying.Yes, but the chance of having everyone else change that is pretty slim.> > Apparently if someone with such a client replies to me, sup will > > automatically reply to the person, not to the list. > > This might (and I really am guessing here) be because sup skips > duplicates (by message id). In the first exaple you get the mail > twice (assuming you are subscribed to lista). The one you most > probably get first (and gets indexed first) is the message you get > directly and which does not have list headers. The one which makes > the extra hop via list server (and has the list headers) gets to you > later and is simply discarded as duplicate. When you go reply the > mail, sup is perfectly unaware of it being a list mail.That makes sense. Could it check for the presence of list headers and discard the one without? One thing I noticed with the list in CC mails is that replying to ''recipient'' would reply to the list. I don''t get that from a semantic point.> > It''s really annoying to send to single people instead of the list by > > accident. > > I can tell you it is _way_ more annoying to accidentally send private > mail to list (I''ve done that, once _really_ badly). So this behavior > (automatical selection between list reply and private reply) should > IMO be conservative in this sense.I see your point. I don''t know what conservative behavior would be in this case. I think consistency is key though. -- Philipp -- "Wir stehen selbst entt?uscht und sehn betroffen / Den Vorhang zu und alle Fragen offen." Bertolt Brecht, Der gute Mensch von Sezuan
Philipp ?berbacher, 2010-08-19 15:40:> I''d like it the other way around, always reply to list and simply > change to ''sender'' in the (at least for me) rare cases where it''s > not supposed to go to the list.It is reasonable that replying to list mail defaults to "Mailing list" as reply mode. Sup does pretty good job being list aware. I think the problem here is sup failing to detect a list mail and treating it as "ordinary" mail. There are at least two cases when sup could (afaik) miss a list mail. 1) A mail which is sent to/cc/bcc a list member and the list so that list member gets two copies of the mail. Sup "sees" only the first one to arrive, which most probably is the non-list duplicate. 2) A similar mail than previous but the copy via list is not delivered at all. Maybe the list performs delivery magic ("whoa, cc-header shows that this list member already got the mail) or whatelse. You can enumerate mailing lists to mutt and have it consider mails to/cc a such address as list mails regardless of the precence of list headers. -- Tero Tilus ## 050 3635 235 ## http://tero.tilus.net/
Excerpts from Tero Tilus''s message of Thu Aug 19 15:07:37 +0200 2010:> I think the problem here is sup failing to detect a list mail and treating it > as "ordinary" mail. There are at least two cases when sup could (afaik) miss > a list mail. > > 1) A mail which is sent to/cc/bcc a list member and the list so that list > member gets two copies of the mail. Sup "sees" only the first one to arrive, > which most probably is the non-list duplicate.If sup is doing the removal of duplicates, could we get sup to update the List-* headers if there are any new ones? -- 77660
Excerpts from Tero Tilus''s message of Thu Aug 19 08:49:21 -0400 2010:> Whoa! > > Mailing lists _can_ do that, but they should not. Ironically you > happend to mention one of the reasons why they should not. :D > > Let''s consider the max damage done by modifying vs. not modifying > reply-to: If the header is unmangled the user may in certain > sircumstances accidentally send private mail instead of list mail. > That''s an annoyance at most. If the header is mangled by list server > the user may in certain sircumstances accidentally send list mail > instead of private mail. That may cost the user a lot of goodwill, > money or her job. Ergo, list setting reply-to header violates the > principle of least damage.Yes, that''s one of the endless philosophical debates about email and mailing lists. I do actually agree with that, it is better not to set the Reply-To header. I was using as an example to show that Sup is behaving correctly in this case :-). Cheers, Israel
Excerpts from Philipp ?berbacher''s message of Thu Aug 19 08:58:38 -0400 2010:> That makes sense. Could it check for the presence of list headers and > discard the one without?Are you using GMail? If you are, it is not possible. GMail also discards duplicated messages, so if you get first the message delivered directly to you, you will not see the list message ever. Cheers, Israel
Excerpts from Israel Herraiz''s message of 2010-08-19 15:41:33 +0200:> Excerpts from Philipp ?berbacher''s message of Thu Aug 19 08:58:38 -0400 2010: > > That makes sense. Could it check for the presence of list headers and > > discard the one without? > > Are you using GMail? If you are, it is not possible. > > GMail also discards duplicated messages, so if you get first the > message delivered directly to you, you will not see the list message > ever. > > Cheers, > IsraelNope, I''m not using gmail at all, but thanks. I checked my mail providers options and found nothing with regards to deleting duplicates. Checking all the mailinglists settings will take a while though. -- Philipp -- "Wir stehen selbst entt?uscht und sehn betroffen / Den Vorhang zu und alle Fragen offen." Bertolt Brecht, Der gute Mensch von Sezuan
Excerpts from Tero Tilus''s message of 2010-08-19 15:07:37 +0200:> Philipp ?berbacher, 2010-08-19 15:40: > > I''d like it the other way around, always reply to list and simply > > change to ''sender'' in the (at least for me) rare cases where it''s > > not supposed to go to the list. > > It is reasonable that replying to list mail defaults to "Mailing list" > as reply mode. > > Sup does pretty good job being list aware. I think the problem here > is sup failing to detect a list mail and treating it as "ordinary" > mail. There are at least two cases when sup could (afaik) miss a list mail.Yep, sup is definitely doing a good job, it seems to be sane in almost any case.> 1) A mail which is sent to/cc/bcc a list member and the list so that > list member gets two copies of the mail. Sup "sees" only the first > one to arrive, which most probably is the non-list duplicate. > > 2) A similar mail than previous but the copy via list is not delivered > at all. Maybe the list performs delivery magic ("whoa, cc-header > shows that this list member already got the mail) or whatelse.I just checked my most active lists and they all were set to ''avoid duplicates if your address is in CC or TO''. I changed that and will see what happens. Maybe sup is clever enough to discard the non-list message.> You can enumerate mailing lists to mutt and have it consider mails > to/cc a such address as list mails regardless of the precence of list > headers.I don''t know anything about mutt and don''t consider using it. I''m fairly happy with sup :) -- Philipp -- "Wir stehen selbst entt?uscht und sehn betroffen / Den Vorhang zu und alle Fragen offen." Bertolt Brecht, Der gute Mensch von Sezuan
Philipp ?berbacher, 2010-08-19 17:26:>> You can enumerate mailing lists to mutt and have it consider mails >> to/cc a such address as list mails regardless of the precence of list >> headers. > > I don''t know anything about mutt and don''t consider using it. I''m fairly > happy with sup :)I wasn''t suggesting you to go mutt yourself. ;) I was just wondering how other mailclients have resolved this problem of duplicate replies (one list and one off-list) from ill-behaving mailclients. Possible solution would be to maintain list of known mailinglist addresses (say mailinglists.txt, just like we now have labels.txt) and consider all mails targeted to a list address as list mail even if they wouldn''t have list headers. Any considerations on the implications that would have? -- Tero Tilus ## 050 3635 235 ## http://tero.tilus.net/
Helge Titlestad, 2010-08-19 16:21:> If sup is doing the removal of duplicates, could we get sup to > update the List-* headers if there are any new ones?That would help with the off-list dupe getting indexed first, but not with the (pretty common) mailinglist setup where the list server plays clever and skips delivery to mailinglist members already appearing in recipient list. -- Tero Tilus ## 050 3635 235 ## http://tero.tilus.net/
Excerpts from Tero Tilus''s message of 2010-08-19 20:24:18 +0200:> Philipp ?berbacher, 2010-08-19 17:26: > >> You can enumerate mailing lists to mutt and have it consider mails > >> to/cc a such address as list mails regardless of the precence of list > >> headers. > > > > I don''t know anything about mutt and don''t consider using it. I''m fairly > > happy with sup :) > > I wasn''t suggesting you to go mutt yourself. ;) I was just wondering > how other mailclients have resolved this problem of duplicate replies > (one list and one off-list) from ill-behaving mailclients.My guess is that some client use primitive means like the subject, but maybe there''s some more header information that''s useful, like reference or in-reply-to. Again, I don''t know much about headers.> Possible solution would be to maintain list of known mailinglist > addresses (say mailinglists.txt, just like we now have labels.txt) and > consider all mails targeted to a list address as list mail even if > they wouldn''t have list headers. > > Any considerations on the implications that would have?So it would compare TO and CC with the list, and if it finds it, treats it as list mail. Well, I guess for some of the ''treat as list mail'' things it might need some headers, but I really don''t know. -- Philipp -- "Wir stehen selbst entt?uscht und sehn betroffen / Den Vorhang zu und alle Fragen offen." Bertolt Brecht, Der gute Mensch von Sezuan
Sort of an edge-case, but what would this mean for cross-posted messages, i.e., messages sent to multiple mailing lists? ~d Quoting Tero Tilus <tero at tilus.net>:> Philipp ?berbacher, 2010-08-19 17:26: >>> You can enumerate mailing lists to mutt and have it consider mails >>> to/cc a such address as list mails regardless of the precence of list >>> headers. >> >> I don''t know anything about mutt and don''t consider using it. I''m fairly >> happy with sup :) > > I wasn''t suggesting you to go mutt yourself. ;) I was just wondering > how other mailclients have resolved this problem of duplicate replies > (one list and one off-list) from ill-behaving mailclients. > > Possible solution would be to maintain list of known mailinglist > addresses (say mailinglists.txt, just like we now have labels.txt) and > consider all mails targeted to a list address as list mail even if > they wouldn''t have list headers. > > Any considerations on the implications that would have? > > -- > Tero Tilus ## 050 3635 235 ## http://tero.tilus.net/ > _______________________________________________ > sup-talk mailing list > sup-talk at rubyforge.org > http://rubyforge.org/mailman/listinfo/sup-talk >