ncurses mouse api: crashes if not initialized about pdcurses

[wmcbrine]

Personally, I'd be OK with it crashing, since that tells you right away "oops, forgot initscr()". But, fine, I realize curses doesn't work that way in general. I've put in SP checks throughout mouse.c (to be committed later).

For this particular function, though, I'm not really seeing the logic of returning the maximum interval on no SP. I realize that's what ncurses does, but not why. Zero would seem to make at least as much sense, reflecting the fact that you're not going to get any click events (or any other input events).

[GitMensch]

I'm not really seeing the logic of returning the maximum interval on no SP

Maybe that's a "proper way" to know the maximum up-front? I don't know, it was just something I've stumbled over.

[wmcbrine]

For now, I've decided to return ERR in this situation. 7e04826

[Bill-Gray]

@wmcbrine 'fixed' this by having a slew of functions return ERR if called before initscr(). However, I agree strongly that (in most cases) the program should crash, and certainly should do so when compiled for debug mode. Calling, for example, init_color() in ncurses without init_scr() does not crash, but it also has no effect. If you do it, you've done something wrong, and could spend some time trying to figure it out if your mistake is simply ignored.
For my fork, I've added an assert(SP) for each such case. This -- at least in debug mode -- makes considerably more sense to me. I also revised mouseinterval() to follow ncurses' lead and return the maximum mouse interval (which I made a defined constant and bumped up to 32767 milliseconds).

[wmcbrine]

Returning ERR is the documented behavior for most of these functions, per both X/Open and the ncurses man pages. But certainly, more detailed feedback would be desirable in some cases.