Re: C-x d: don't make user have to retype misspelled dirs

[Kevin Rodgers]

Dan Jacobson wrote:

> Dan>     er, um, Checkmate: how come C-x C-r in the same situation will not let
> Dan>     the user get away with a RET here, and instead does the comfy cosy
> Dan>     coaxing thing, whereas C-x d does the parental punishment thing.
>
> RMS> I don't follow what you are saying about C-x C-r.
> RMS> Would you please be more explicit?
>
> $ ls -d /var/lo*
> /var/local  /var/lock  /var/log  /var/lost+found
>
> C-x d / v a r / l o <return> C-x C-r / v a r / l o <return> <return>
>
> Consider the treatment one gets on the first <return> vs. the second
> and third.  The former is like tough luck bub, whereas the latter is
> VIP helping environment warm & fuzzy.

I believe this all comes back to the fact that there is only a single
`interactive' character code for reading a directory ("D"), and so
there isn't a way to distinguish between an existing directory and a
possibly nonexistent directory (cf. "f" and "F" for files).  Neither
is there a read-directory-name analogue for read-file-name, which
accepts an optional MUSTMATCH argument.

--
Kevin Rodgers <kevinr@ihs.com>