Re: conf-space-mode

[Wolfgang Jenkner <wjenkner@inode.at>]

Richard Stallman <rms@gnu.org> writes:

>     The way you state this suggests that perhaps you forgot that I
>     actually sent a patch?
>
> I saw your patch but decided to fix the problem a different way.

Not very different, though.

>     You didn't address the problematic use of `current-prefix-arg'.
>
> Yes I did, in the second change.

I'm afraid we are not talking of the same problem here.

I think what you call the `second change' is making C-u C-c SPC work.

I meant the part where I remarked that, e.g., `find-file-at-point'
uses `current-prefix-arg'.  A call to `conf-space-mode' while
'find-file-at-point' is being executed may still see the same prefix
argument which was passed to `find-file-at-point'.  I gave a rather
surprising example of this situation, where the user might not be
aware of some config file being inserted in a buffer.  That example is
slightly bogus however since it actually works correctly with
'ffap-bindings'.  Here is something that should ahem... not work:

(ffap-bindings)
(setq ffap-require-prefix t)

Place point after the colon of ftp.foo.bar: and do C-u C-x C-f

My patch makes `conf-space-mode' heed the prefix argument only if it
is the actual command being executed.  I think this is enough since
you can always visit a config file and then do C-u C-c SPC to adjust
`conf-space-keywords'.  But then again, I never use this feature.

> The first one fixes the other problem you mentioned, that of asking
> for confirmation twice.  You used the words "by design" and I took
> that to mean you thought this was a feature, so I assumed you had
> not tried to change this.

You assumed?

>  But I considered it a bug.  So the first thing I did was fix this.

The imenu stuff is slightly broken now (given my test file x, try C-x
C-f x <return> y and observe that `imenu-generic-expression' has two
"Parameters" entries).