MAINT@DERRZE1.BITNET (Rainer M Woitok) (01/16/90)
While I'm continuing to utilize the new features of MAIL/MAILBOOK 89-02- 0A, I found two other things I'd like to see: 1. When you display the columns SUBJECT, DATE and FROM in this order in the menu, there's no blank between the date and the From- information. This looks a bit ugly. In fact, in the standard display (FROM, TO, DATE and SUBJECT), there isn't a blank between date and subject either, but since the subject starts with either a blank or an asterisk, there is sort of a separation between these two. 2. When you sort according to subjects, the resulting sequence will be eg "*A", " A", "*B", " B" etc., which is quite ok. Now consider, subject "A" came up twice, ie your notebook contains messages " A" and "*A" dating from yesterday and today respectively, and messages " A" and "*A" dating a month back. Then sorting for subject and date would yield "*A", "*A", " A", " A" and not "*A", " A", "*A", " A" which would be rather more useful. Could this easily be changed? Sincerely Rainer .----------------------------------------------------------------------. | Rainer M. Woitok | Phone : (+49 9131) 85-7811,-7031 | | Regionales Rechenzentrum | Fax : (+49 9131) 30 29 41 | | Friedrich-Alexander-Universitaet | Telex : d 629 755 tf erl | | D-8520 Erlangen | e-Mail : MAINT@DERRZE1.BITNET | | West Germany | | '----------------------------------------------------------------------'
NJG@CORNELLA.BITNET (Nick Gimbrone) (01/17/90)
On Tue, 16 Jan 90 14:27:09 CET Rainer M Woitok said: >2. When you sort according to subjects, the resulting sequence will be > eg "*A", " A", "*B", " B" etc., which is quite ok. Now consider, I'm not sure I agree about this being ok. Since "*A" and " A" have the same subject, but one is/should be written after the other I'd rather that the "*" or " " column not be included in the sort and that the GMT date/time field be used as a secondary sort field (such that " A" always preceeded "*A"). I've not verified how the code is written/commented to work :-).
SCHAFER@RICEVM1.RICE.EDU (Richard A. Schafer) (01/17/90)
On Tue, 16 Jan 90 17:55:56 EST Nick Gimbrone said: >On Tue, 16 Jan 90 14:27:09 CET Rainer M Woitok said: >>2. When you sort according to subjects, the resulting sequence will be >> eg "*A", " A", "*B", " B" etc., which is quite ok. Now consider, >I'm not sure I agree about this being ok. Since "*A" and " A" have the >same subject, but one is/should be written after the other I'd rather >that the "*" or " " column not be included in the sort and that the >GMT date/time field be used as a secondary sort field (such that " A" >always preceeded "*A"). >I've not verified how the code is written/commented to work :-). Not only do I believe that this is *not* OK, as far as I can determine that's not the way the code actually works. As far as I can verify, " A" is always sorted before "*A" for precisely the reason that you want replies to follow the original. If it doesn't work that way (and so far as I can tell it does), it's a bug. Richard
MICHAEL@UTORONTO.BITNET (Michael Wagner) (01/18/90)
>I'm not sure I agree about this being ok. Since "*A" and " A" have the >same subject, but one is/should be written after the other I'd rather >that the "*" or " " column not be included in the sort and that the >GMT date/time field be used as a secondary sort field (such that " A" >always preceeded "*A"). Since this is wishlist time, I'd rather have the messages sorted into "thread" order, i.e. use the string of "in-reply-to" message-ids where available, and only use the GMD time/date logic when in-reply-to is not available or not tracable. This would also alert you to messages that were "orphans" Michael
MICHAEL@UTORONTO.BITNET (Michael Wagner) (01/18/90)
>while I would prefer after sorting according to subject and date in >ascending order to have them in the sequence shown first. I think, with all due respects, that we cannot force the subject and date fields into a duty to which they were not intended, without a lot of 'stand on your head' code. Lets step back and find out what we really are trying to do. I would guess that you want to be able to read the messages in the order that corresponds to which messages are replies to which. The best way of doing this is to use the 'in-reply-to' fields, where available, and to use time/date/subject/timezone/cokebottle heuristics only when better information isn't available. There are, of course, User Agents that don't produce standard In-Reply-To fields, so this won't always work, but part of the reason for the lack of standardization is that no one has cared up to this point. If some software actually used the field for something, there would be impetus for people to clean up their act. The problems with using time and subject are: times aren't set reliably in many environments, and subject names get reused. These are, in my opinion, substantial problems that will not be solved by us. >I hope, this time at least someone is recognizing my point, even if she >or he doesn't agree. Did I understand you correctly? Michael >Sincerely > Rainer
MAINT@DERRZE1.BITNET (Rainer M Woitok) (01/18/90)
On Tue, 16 Jan 90 17:43:34 CST Richard A. Schafer said: >On Tue, 16 Jan 90 17:55:56 EST Nick Gimbrone said: >>On Tue, 16 Jan 90 14:27:09 CET Rainer M Woitok said: >>>2. When you sort according to subjects, the resulting sequence will be >>> eg "*A", " A", "*B", " B" etc., which is quite ok. >>I'm not sure I agree about this being ok. >> ... >Not only do I believe that this is *not* OK, as far as I can determine >that's not the way the code actually works. As far as I can verify, >" A" is always sorted before "*A" for precisely the reason that you >want replies to follow the original. If it doesn't work that way >(and so far as I can tell it does), it's a bug. Sorry, I think, I just didn't get my point accross (I didn't mention whether I was using SORTUP or SORTDOWN). But my point was not whether or not subject " A" should precede "*A", but the following: Assume a notebook containing the four items subject " A", date 88-01-01 subject "*A", date 88-01-02 subject " A", date 89-01-01 subject "*A", date 89-01-02 Then SORTUP SUBJECT DATE (as well as SORTUP SUBJECT alone) will produce the sequence subject " A", date 88-01-01 subject " A", date 89-01-01 subject "*A", date 88-01-02 subject "*A", date 89-01-02 while I would prefer after sorting according to subject and date in ascending order to have them in the sequence shown first. I hope, this time at least someone is recognizing my point, even if she or he doesn't agree. Sincerely Rainer .----------------------------------------------------------------------. | Rainer M. Woitok | Phone : (+49 9131) 85-7811,-7031 | | Regionales Rechenzentrum | Fax : (+49 9131) 30 29 41 | | Friedrich-Alexander-Universitaet | Telex : d 629 755 tf erl | | D-8520 Erlangen | e-Mail : MAINT@DERRZE1.BITNET | | West Germany | | '----------------------------------------------------------------------'
MAINT@DERRZE1.BITNET (Rainer M Woitok) (01/18/90)
On Wed, 17 Jan 90 12:08:22 EST Michael Wagner said: > ... >I would guess that you want to be able to read the messages in the >order that corresponds to which messages are replies to which. The >best way of doing this is to use the 'in-reply-to' fields, where >available, and to use time/date/subject/timezone/cokebottle heuristics >only when better information isn't available. This idea is clearly supperior to what I suggested, though, I fear, a lot more difficult to implement. The solution of disregarding the first character ("*" or " ") of the subject field whenever an additional sort field was specified would be a quick (& dirty), though not perfect one. Besides you had to do that anyway, when no 'In-Reply-To:' header field was available. >Did I understand you correctly? Michael Yes, I think you did. Sincerely Rainer .----------------------------------------------------------------------. | Rainer M. Woitok | Phone : (+49 9131) 85-7811,-7031 | | Regionales Rechenzentrum | Fax : (+49 9131) 30 29 41 | | Friedrich-Alexander-Universitaet | Telex : d 629 755 tf erl | | D-8520 Erlangen | e-Mail : MAINT@DERRZE1.BITNET | | West Germany | | '----------------------------------------------------------------------'
NADWL@HDEDH1.BITNET (Scott Ophof) (01/18/90)
On Wed, 17 Jan 90 11:21:41 EST Michael Wagner said: >>I'm not sure I agree about this being ok. Since "*A" and " A" have the >>... >Since this is wishlist time, I'd rather have the messages sorted into >"thread" order, i.e. use the string of "in-reply-to" message-ids >where available, and only use the GMD time/date logic when >in-reply-to is not available or not tracable. This would also >alert you to messages that were "orphans" I second this wish/motion, and have a feeling that *effectively*, Rainer Woitok means the same thing. Please correct me if I'm wrong, Rainer. Regards. Scott/
NADWL@HDEDH1.BITNET (Scott Ophof) (01/19/90)
On Wed, 17 Jan 90 12:08:22 EST Michael Wagner said: >>while I would prefer after sorting according to subject and date in >>ascending order to have them in the sequence shown first. >... >I would guess that you want to be able to read the messages in the >order that corresponds to which messages are replies to which. The >... Would something like: "Message-id: nnnn (follows msgid mmmm)" do this? Regards. Scott/
MICHAEL@UTORONTO.BITNET (Michael Wagner) (02/03/90)
>Re using the chain of Message-Ids to sequence messages: > >I'm at a loss to see how this could possibly be implemented in the context of >a discussion list such as this. Postings to a discussion list do not occur in >a linear sequence. There is not one single series of message-reply-reply...; >rather, the structure is a tree form, with multiple replies possible to any >given message. How would you sequence two messages with the same message-id >in the In-Reply-To: header line? At least you would want a secondary value on >which to sort - the date, perhaps? Of course it's a tree. Trees have an order. It's just not a simple order, and it isn't clear that you can just pass it off to the SORT command. But it's a very useful order. The usual order for this sort of thing is a depth-first tree walk, with some other key (fancy time-zone compensated date, in this case) resolving the order of the siblings. But, to tell you the truth, I don't want to see them on the front screen in this order, so I don't think the linear representation is much of an issue. I would rather see them clumped on the front screen as MAILBOOK 4 items HELP! My help contains no help MAILBOOK 18 items More ideas for the wish list MAILBOOK 2001 items More congratulations for Richard on a job well done This allows me to find my personal mail in amongst the discussion group mail, and allows me to skip the 2001 thank yous to Richard with a single keystroke (I echo the thank you sentiment, mind you. I'm just tired of reading it over and over again). More to the point, when I am looking at a piece of mail, I want 3 PFKeys: PARENT, SIBLING, and CHILD. These should use the message-id information when it is available, and only resort to date/time/timezone/cokebottle heuristics when more reliable means aren't available. PARENT and CHILD are a lot like PREVIOUS and NEXT, once you have figured out how to sort your mail correctly (which I never have). SIBLING doesn't seem to really exist in MAIL/BOOK. For those who did not study computer science in English or are not familiar with this terminology of trees, here is a diagram: Original Note If you are looking at Reply A, | the PARENT key gets you the | original note, the SIBLING +---------+----------+------------+ key gets you R B, R C, & R D, and | | | | the CHILD key gets you R R A. Reply A R B R C R D | R R A By the way, I think these names for the keys would not be good end-user names for the keys. >- Lawrence Michael
NADWL@HDEDH1.BITNET (Scott Ophof) (02/06/90)
On Fri, 2 Feb 90 11:15:59 EST Michael Wagner said: >>Re using the chain of Message-Ids to sequence messages: >>... >MAILBOOK 4 items HELP! My help contains no help >MAILBOOK 18 items More ideas for the wish list >MAILBOOK 2001 items More congratulations for Richard on a job well done I second the motion for this type of display (as an option, of course). >More to the point, when I am looking at a piece of mail, I want 3 PFKeys: >PARENT, SIBLING, and CHILD. These should use the message-id information >when it is available, and only resort to date/time/timezone/cokebottle >heuristics when more reliable means aren't available. Another second here. Especially the SIBLING idea (I'm running out of PF keys, by the way...). PF keys sensitive to the cursor-position? Generally, I think it's time to evaluate what's really MAIL/MAILBOOK stuff, and what is nice & handy stuff (but not really necessary), and putting all those extra's in a separate macro; might make MAILB00K XEDIT easier to manage, and maybe give users a better place to put their own ideas? I know about MAILUSER XEDIT, but I don't feel that to be the right place. Regards. Scott/