Comment 6 for bug 82708

:execute "set t_kP=\<Esc>[5;*~"

etc, may be used in lieu of my versions, to avoid literal escapes in the vimrc.

> I wonder why vim assumes to have anything other than an a vt100 compat term if it has a non-specified TERM var

Not sure what you mean: "screen" isn't a non-specified term, and it has terminfo capabilities. And, vim does in fact assume vt100 (ansi) if TERM isn't set. But, as you've already said, vt100 has no notion of Control-PageUp. The real problem is that there's just no way to specify, using terminfo or termcap, what Control-PageUp will look like in terms of control sequences.

The above isn't /precisely/ true: AFAICT, vim would recognize its special "*" notation even if it came from terminfo; it's just that using that syntax would then break pretty much any other terminal program's ability to recognize (regular) PageUps.

> on a philosophical note, should a bug ever be rejected just because no one can think of a fix?

*sheepish grin*

I'm not sure about that. You are certainly welcome to reopen this bug, if you'd like. I mostly closed it because I figured you'd be okay with doing that, given my explanation. I do think that, in a case where a fix is truly impossible in every sense of the word, closing the bug is entirely appropriate.

But in this case, a fix is, technically, not completely impossible; just very involved. The right (and AFAICT, only) fix for this would be to (1) enhance ncurses/terminfo to provide a way to specify how various special keys should look if they have modifiers; (2) enhance screen to support this mechanism and process/translate modifier/special-key combos; and (3) enhance vim to understand and support this mechanism. So, rather than open this general bug, perhaps opening these three bugs (consecutively, not simultaneously) would be preferable.

This really needs to happen at some point—or, if not this, I would love to see terminfo at least provide support for xterm-style or gpm-style mouse tracking (I just dealt with a similar bug involving vim's inability to detect mouse support in screen: fortunately, that one was fixable, but the ideal solution would still have been in terminfo).

I'll tell you what: let's leave this bug closed, and I'll file a bug against ncurses, and get some discussion going upstream. If I don't leave a comment on this bug pointing to the new bug within a reasonable amount of time, reopen it. :)