Beartooth posted on Sat, 25 Dec 2010 16:45:17 +0000 as excerpted: > On Wed, 22 Dec 2010 06:14:12 +0000, Duncan wrote: > [....] >> Similarly, it's no big deal to tell pan to watch your own posts, or >> those of any other author, since that's a single entry in the score >> file. [....] > > OK, I'll bite. What are the fine points of how I do that? (I mean > additions, colors, and whatever I haven't thought of yet.) Let me > backtrack. > > I knew that scoring has something to do with watching, and how to > spell it. I can guess that you mean the file named "Score" in .pan2, but > that daunted me.
"Watching" is nothing more than "scoring" to +9999 (or above). Similarly, "ignoring" is nothing more than scoring to -9999 (or below). The entries for each (watching/ignoring) are simply shortcuts that create scores of +/-9999 without having to go thru the score dialog. > I thought to try looking through the list of choices when I right- > clicked on one of my own posts, and found "Add a scoring rule." > > Following my nose from there, I started playing with adding > numbers to my score -- discovering quite soon that I got a variety of > highlighting colors, and that (as I should have anticipated) Pan > interpreted all my obsolete addresses as different authors. (I did half- > recall, half re-find Colors under Add Preferences.) > > That was enough, especially when done after sorting by author, to > pick up most -- and did hit threads which I didn't start, but had > participated in. The last was a real help. > > So what I still don't know is whether there are still more tricks > to this, such as optimal numbers (or numbers which a large plurality of > users find optimal in practice) to add to particular authors, subjects, > or whatever. And I don't see what most of the (doubtless highly > practical) elements you can score by are good for. A long time ago, before pan had scoring, it had filtering. You could still watch or ignore, but it didn't score back then, it simply filtered. Filtering is binary, or rather, as pan had it implemented, trinary (watch, normal, hide, nothing in-between). Scoring does sort of the same thing, but in a "fuzzy logic" way, by giving you a whole range of choices and allowing multiple factors to apply, adding a few points here due to one scoring factor, subtracting a few there due to another, with pan displaying the final result as a color-coded score. As mentioned above, it's still possible to watch or ignore, but those are now implemented simply as "really high" and "really low" scores, +/-9999. Scores work in parallel with color-scoring, with colors set as you discovered on the colors tab of the preferences dialog. If you take a look at that... (One thing I like about pan is that you can open the dialogs and still work with main pan and the posting window -- it doesn't block you until you close the window, as some apps do; this is very useful when referring to them as I compose a post... or as you read my composition. =:^) As I was saying, if you take a look at the colors tab, you'll note the five scoring color-zones/levels, two below zero, three above, making six levels total if you include normal/zero as a level. On the view, header-pane sub-menu, it's possible to match (or not) any of these same scoring levels/zones, so technically, a particular post's score and whether that scoring zone/level is "ignored" (as in, not displayed, as opposed to the specific ignored level -9999 or below) or "watched" (as in, displayed with a color-hilite, as opposed to the specific watched level, +9999 or above), is now entirely arbitrary -- you decide with the color and view settings what levels are "watched" or "ignored", regardless of whether they're actually in the specifically labeled watched or ignored levels. It's worth noting while we're talking about the view, header-pane sub- menu, that it's ALSO possible to set view-matched (only), view-matched- threads (the entire thread if any post matches), or view-matched-subthreads (anything below a matching post). This sometimes mixes people up as they change the view in other ways but forget this, and don't get the behavior they expect. It's also worth noting that show matching subthreads, set along with viewing a particular score level, will allow you to do what you originally requested, view only the threads/subthreads to which you've posted, provided that you've set scores for your own posts appropriately **OR** use the "match my posts" option (which I had forgotten about). However, because that ALSO toggles the thread/subthread view for read/unread, cached/uncached, and complete/incomplete, it might not have the expected effect. This is particularly true if like me you normally view only unread posts, and sometimes set posts unread to read again later, but don't want the entire subthread showing up just because a parent is marked unread. For this reason I tend to leave it set to show-matched-articles only, and forgot, in my earlier post, that this feature combined with the match my own posts (or with scoring my own posts) feature would allow you to do what you requested. But I know you like really simple instructions and fear you might get mixed up trying to track the various combinations, so perhaps it's better that you, like me, simply leave it set to show-matched-articles (only). Meanwhile, back to scoring rule variants and what they might be good for... Note that in the scoring rule dialog, in addition to setting the score's absolute /value/ (either picking your own value, or using the arbitrary watched/ignored +/-9999 values), it's possible to set it to increase or decrease a score by a /relative/ amount, the default being 100 points. If a score is set to an absolute value, that's what pan sets it to, and it stops processing more scores, since that's an absolute value. But if it's set relative... the scoring logic will continue to try to match additional scores, adding and subtracting any relative score that matches, until a final score is established. This is where the fuzzy logic comes into play. Many people don't like excessive cross-posting, so they might score articles down a few points for that. They like certain authors, but not enough to watch them, so they score them up a few points. In text groups they might not want to see binaries, so score articles down that are more than 200 lines or 14 KB. Conversely, in many binary groups, text posts and really small binaries are usually spam, so there, they'd score such articles down by some number of points. Then one of these authors they like comes along and posts a series of small binaries to a binary group the scoring user follows, cross-posted to just one other related group. If the scoring user has his relative scores set correctly, cross-posting to just one other group will be only a small negative (or none), as will be the fact that the series happens to be small binaries. The score for liking the author will counteract these two negatives, so the overall score will still be a bit positive and the scoring user will see the post. But if the spammer relative of this liked author gets ahold of the account and tries to spam to a dozen groups at a time, the higher relative negative score of cross-posting to a dozen groups at once will, if the relative scores have been set well, cancel out the bonus from the liked author relative score, with the result being a much more negative score that the scoring user won't see at all. Of course, if you're SURE you like or dislike an author's posts strongly enough, instead of setting a relative score, you'll set an absolute score, and it'll always apply (as long as no earlier absolute score applies first, pan will stop processing when it sees one and apply that score). Then all the relative scores you've set won't affect posts by this author, which will always be scored the same, by the absolute score. This is (IIRC) what pan's watched and ignored functions do, set an absolute score of very high or very low (+/-9999). That covers most of the scoring criteria and demonstrates the flexibility of relative scoring. I didn't mention subject, but the value of it for scoring should be obvious. That leaves age, and references. References, as in the message-ids in the references header, is what is used for the watch/ignore thread/subthread functionality as I mentioned in the previous post, and of course you can use it for general scoring as well. Age... well, that lets you score by age. If you don't expire messages or at least don't expire them right away, and a thread goes on for some time, that lets you score (which means color-code, since that's what scores do) by age. (Back in old-pan, expiry was based on the server, not on a local setting as it is now. If the server kept messages for a long time or didn't expire them at all, scoring by age, along with rules based on that score, new-pan doesn't have rules, unfortunately, but the combination of age-scoring and rules could be used to expire messages before they'd normally expire on the server. Obviously, pan's expiry method is set locally now, and we don't have rules, so age-scoring isn't useful for expiry as it was then, but scoring by age remains an option.) > I recall talk of a FAQ, and there is a thread from 2007, but no > link from the Pan home page to a manual .... There was someone putting together a manual, or readme or something, back then, based partly on posts from this list/group. I don't know what happened to that project... -- Duncan - List replies preferred. No HTML msgs. "Every nonfree program has a lord, a master -- and if you use the program, he is your master." Richard Stallman _______________________________________________ Pan-users mailing list Pan-users@nongnu.org http://lists.nongnu.org/mailman/listinfo/pan-users
