On Fri, Apr 27, 2007 at 01:17:39PM +0200, A.J.Mechelynck wrote: [...] > >Aha, I think, that here we talk about a bug! If I hit Ctrl-N, get the > >needed completion, and then start typing further, then the completion > >stops as soon as there are no more matches. BUT. If after Ctrl-N I hit > >Backspace first, and then continue typing, then completion does not stop > >even when there are no more matches. In this case only Esc or Ctrl-Y > >aborts the completion. > > > > Ah, finally, I can reproduce that. I think it's only a missing screen > redraw. If I backspace, then type a few letters, then (let's say) a period, > the message "match m of n" is still there. Hitting <Left> makes it > disappear (replaced by "-- INSERT --"), after which <Right> brings the > cursor back where it belongs.
Unfortunately it's not just a missing screen. If you try to do completion again it won't work. So again, type the beginning of word, hit Ctrl-N, hit Backspace, type ( and beginning of another word, hit Ctrl-N to complete it. It won't work saying that there are no matches. That's because old completion is still active and it tries to complete the whole thing. This happens very often when one has to write C functions :) So, I would say this is a more serious bug than just missing update of the screen. Actually, I've hit another bug few times already, unfortunately I can't reproduce it. When I type quickly and miss a key, then I try to undo my changes and instead get back the message "Can't undo, the line numbers are incorrect" or something like that. Usually it happens after I hit 2-3 keys wrong, so I still can't reproduce it. Well, when I type japanese text in "over the spot" mode this is "normal". But everyone decided that this is a bug in the japanese input. Looks like it can be triggered also in non-japanes input. -- Minds, like parachutes, function best when open
