[9fans] du

[9fans] du

[Lucio De Re]

Plan 9 du(1) is inconsistent with BSD's: du -s walks down one
directory in Plan 9 before the summary.  I believe that I can achieve
that effect, should I want it, by entering du -s *, whereas I can't
achieve the same effect as NetBSD's du -s / using Plan 9's semantics.

Do I need a new du of my own or would the community accept a modified
du(1) with BSD semantics?  Is there a reason for the Plan 9 semantics
that I am unaware of?

++L

Re: [9fans] du

[bs]

"rob pike, esq." wrote:

Any history on the "esq"?

thx

Re: [9fans] du

[Lucio De Re]

On Thu, Mar 21, 2002 at 03:44:27PM +0200, Lucio De Re wrote:
>
> No scripts that will break because of the change?  Du(1) is used quite
> extensively because of find's absence, I'd hate to be the one to
> create a problem by trying to be helful :-)
> 
And I tried to mail Rob exclusively, but Mutt obviously makes its own
decisions, so Rob got a duplicate message :-(

I guess I could have looked a little harder...

++L

Re: [9fans] du

[Lucio De Re]

On Thu, Mar 21, 2002 at 08:37:26AM -0500, rob pike, esq. wrote:
> 
> In any case, it's trivial to change.  Just eliminate the variable
> 'top' altogether from the function du() in /sys/src/cmd/du.c.  I've
> made the change on our system.
> 
No scripts that will break because of the change?  Du(1) is used quite
extensively because of find's absence, I'd hate to be the one to
create a problem by trying to be helful :-)

++L

Re: [9fans] du

[rob pike, esq.]

> Plan 9 du(1) is inconsistent with BSD's: du -s walks down one
> directory in Plan 9 before the summary.  I believe that I can achieve
> that effect, should I want it, by entering du -s *, whereas I can't
> achieve the same effect as NetBSD's du -s / using Plan 9's semantics.
> 
> Do I need a new du of my own or would the community accept a modified
> du(1) with BSD semantics?  Is there a reason for the Plan 9 semantics
> that I am unaware of?

This seems reasonable.  I'm not sure why du works this way; it could
be I was trying to reproduce the behavior of the old `research Unix'
version, or I might have thought it was a good idea at the time, or I
might just have screwed up.  The code goes out of its way to behave
like this.

In any case, it's trivial to change.  Just eliminate the variable
'top' altogether from the function du() in /sys/src/cmd/du.c.  I've
made the change on our system.

-rob