Re: cd

[Benjamin Lindner]

Jaroslav Hajek wrote:
> On Wed, Aug 27, 2008 at 1:07 PM, Levente Torok <address@hidden> wrote:
> > On Wed, Aug 27, 2008 at 9:45 AM, Olaf Till <address@hidden> wrote:
> > > On Tue, Aug 26, 2008 at 05:17:03PM -0400, John W. Eaton wrote:
> > > > On 25-Aug-2008, Levente Torok wrote:
> > > >
> > > > | Hi All,
> > > > |
> > > > | I ran into an other simple compatibility issue:
> > > > | In octave, 'cd' does not return anything however it should (pwd).
> > > >
> > > > Yes, this is a known "problem".  I'd consider a patch.
> > > >
> > > > But one thing that should not be "fixed" is that cd by itself should
> > > > change the current directory to $HOME.  ...
> > > Agree that this should not be "fixed", but wouldn't it be weird then
> > > if the current directory is printed before 'cd' changes to the home
> > > directory?  Olaf
> >
> > If matlab compatilibity is a target, than it should follow what matlab does.
> > In matlab "cd" return the current dir and does not change it!.
> > "cd ~"
> > returns nothing but changes curr dir.
>
> Matlab compatibility matters, but is not an absolute rule. Most of
> Octave users use also other tools than Matlab. When you have such an
> ubiquitous command such as cd, you should consider compatibility in a
> more general sense. "cd" without arguments always meant "cd ~" on
> Unix. Matlab's behaviour is apparently modeled after "cd" in DOS
> (which knew nothing about home directories).
> Though I'm sure that Windoze users are an important group (perhaps
> even most important), I bet that most of them actually do not use
> Windows command line much, while most of Linux and Mac users probably
> do.
> But I'm not sure about this. Perhaps some Windows users could tell us
> whether they care about this matter. If they don't, I'd vote for not
> changing the behaviour.

Speaking as windows user, I do not see a problem here if the "cd" 
command behaves in the standard-unix way.
Windows does not provide a really useful terminal, and obviously never 
put emphasis into it, so win32 behaviour at the console should not be 
the reference IMO.

I personally use win32 command line quite a bit, but to retrieve the 
current working directory it is much less complicated to read the CD 
environment variable than reading the output of a "cd" command, so I 
don't use it.
What I wanted to express is, reading the cwd in win32 console, you would 
not call the "cd" command, so I see no point why the cd command in 
octave should return the cwd *for this reason alone*.

Matlab compatibility is a different matter.
Again, personally, I would not follow matlab's conventions in this case. 
I like the "compatibility in a more general sense" as commented above.

benjamin