Re: [PATCH 1/4] add generic, type aware object chain walker

From: Martin Koegler <mkoegler@auto.tuwien.ac.at>
Date: Feb 25, 2008, 08:21 UTC
Message-ID: <20080225082145.GB20048@auto.tuwien.ac.at>
In-Reply-To: <7vskzhmppo.fsf@gitster.siamese.dyndns.org>

On Sun, Feb 24, 2008 at 11:59:15PM -0800, Junio C Hamano wrote:

Show 8 quoted lines

> mkoegler@auto.tuwien.ac.at (Martin Koegler) writes:
> > How should I define the return value of fsck_walk in the presence of
> > multiple errors?
> 
> Returning error is fine.  That is not what I was talking about.
> 
> I was talking about an early return in the code, that does not
> callback once you find an error.

That is not what I was talking about.

We need to define the return code of fsck_walk in the case of multiple errors. What error (of all possible) should it return?

What about this new behaviour:

Callback return values:
* 0 everything OK
* > 0 error, but continue
* < 0 error, but stop

The return value of fsck_walk would be:
* -1, if fsck_walk discoveres an error (eg. not parseable object)
* first callback return value < 0 (and stops further processing in this case)
* first callback return value > 0 (if not callback return value < 0)
* 0, if no errors are discovered

mfg Martin Kögler