Re: tumme messages in the echo area

[Nick Roberts]

 > > The pretest has not started yet and it makes sense to fix all reported bugs
 > > anyway.
 > 
 > I did not say I had anything against it.

You asked if it was safe to fix this in Emacs CVS now.

 > > I don't use Tumme so it would be good if you can test my changes.
 > 
 > I will.
 > 
 > >  >                                             I am a bit skeptical about
 > >  > the "-c" switch.
 > >
 > > What problem can you see?  If you see a better fix then you should
 > > install it.
 > 
 > The problem with it is that it might now work in all shells. For
 > example, on Windows, cmd.exe (the default "shell") uses /c, not -c. I
 > don't know how other shells handle this.

I see now.  I've fixed this as you suggested with shell-command-switch.

 > > AFAIK shell-command is is for interactive use: hence the message in
 > > minibuffer.
 > 
 > I see now that it is an interactive function, but it accepts optional
 > extra parameters which suggests that one can use it non-interactively
 > too, right? Sorry if I misunderstand the purpose and intent of
 > interactive functions.

An interactive function can be used non-interactively but AFAICS the optional
extra parameters for shell-command control the output/error of the command,
not the minibuffer message.  Call-process is more low level and does (I think)
what is needed.  Why not use it?

 > By the way, I checked the source and found this, which proves my point:
 > 
 >   (call-process shell-file-name nil t nil shell-command-switch command)
 > 
 > Please use the same technique, using `shell-command-switch'.
 > 
 > /Mathias

-- 
Nick                                           http://www.inet.net.nz/~nickrob