On Tuesday, September 11 at 07:37 AM, quoth Mutt:
Conversely, you can't be sure that what the user (or something else) has configured for erase is really the backspace key.
Fair enough, but I don't expect mutt to handle pathological users.I'm suggesting we go with "least surprise". The "erase" character, whatever it is and no matter what key emits it, is the character that triggers "backspace-like" behavior. So when I say "mutt, when I hit backspace, do X", I expect mutt to understand backspace the same way that every other program does: whatever key I hit that happens to produce backspace behavior is the one I'm referring to. If I happen to have configured my terminal such that the 'r' character is the "erase" character (such that if I type "bar" I end up with just "b"), then every program I run in that terminal is essentially treating 'r' as my backspace key. Thus when I tell mutt to bind "backspace" to something, I expect mutt to bind 'r' to that something. It's simply consistent.
And if I've done everything in my power short of writing my own terminfo entry to convince my terminal that ^? is the backspace character, then why must mutt be the only program still stubbornly insisting that ^H is backspace? Vim doesn't use $TERM to override the stty setting. Nor does emacs, nor does libreadline, nor does telnet, xchat, bitchx, or any other terminal-based application. *WHY* is this considered "good" behavior for mutt?
When I say "mutt, backspace means foo", what I'm really saying is "Mutt, when I press the key that behaves as a backspace...". Of course mutt can't make sure that they key I press is really *labelled* "Backspace". On many Apple keyboards, it's actually labelled "Delete" even though for all other intents and purposes it's the backspace key. That's because the <BackSpace> key is a reference to "the key that has the backspace effect". Of *course* mutt can't be prescient and somehow find out whether or not my keyboard even *has* a BackSpace key, nor can it find out for certain what that key actually emits. But it CAN trust that when the user says "backspace" they don't think mutt is omniscient. Simply using whatever happens to have the backspace effect (namely, the "erase" character) makes perfect sense, and is consistent with virtually every other terminal-based program. More to the point, I really don't see how using anything else makes any sense at all.
Currently mutt is extracting this bit of information from the terminfo entry. My point is that this is insufficient because the terminfo may be wrong (buggy) or outdated (superseded). If this was our only recourse, I'd say that's fine, but it's not the only recourse. There's a better way to do it! Mutt can't find out if the terminfo entry is wrong or buggy, but it CAN find out if the terminfo entry has been superseded, and I think it *should*.
~Kyle
--
Reality is that which, when you stop believing in it, doesn't go away.
-- Philip K. Dick
Attachment:
pgpBpF8OH3dco.pgp
Description: PGP signature