Re: Extent endpoints

Hi Stephen, Thanks for moving that to xemacs-beta. I found some time to dig deeper into that issue. I quoted my original message below for people who didn't follow comp.emacs.xemacs (probably most of you ;-) I wrote: You were close with the bug in transpose-chars which is just a wrapper for transpose-subr. This is where the funny behaviour comes from. What transpose-subr does is: 1. delete string following point (thus modifying the end-open extent) 2. insert string preceeding point (placing it outside the extent) 3. insert string deleted in 1. before string inserted in 2. (placing it outside the extent) 4. delete string preceeding point Taking my example above the steps are This is an extent-!-. This is an extent-!- This is an extentt-!- This is an exten-!-t This is an exten.-!-t Wow that's a lot just to swap two strings before and after point ;-) I consider this behaviour buggy because it has totally unexpected results even for someone who *does* now about corner-cases at extent-endpoints. In the example above, two characters that had been snugly inside the extent find themselves outside the extent for no good reason. My statement about the -- still unoffical -- after-change-functions extent property is *not* related to the bug described above. I assume that extent after-change-functions are implemented using the buffer after-change-functions, which would show that behaviour unless extents and their positions are being carefully kept track of. I sent Hrvoje a more detailed email about that, since he apparently implemented that feature. Best regards, Marcus