From: Nicolas Pitre <nico@fluxnic.net>
Date: Mon, 12 Apr 2010 15:47:25 GMT
Subject: Re: [PATCH 2/2] index-pack: rationalize unpack_entry_data()
Message-ID: <alpine.LFD.2.00.1004121145290.7232@xanadu.home>
In-Reply-To: <4BC2BF69.8050100@viscovery.net>

On Mon, 12 Apr 2010, Johannes Sixt wrote:

> Am 4/12/2010 4:57, schrieb Nicolas Pitre:
> > -	for (;;) {
> > -		int ret = git_inflate(&stream, 0);
> > -		use(input_len - stream.avail_in);
> > -		if (stream.total_out == size && ret == Z_STREAM_END)
> > -			break;
> > -		if (ret != Z_OK)
> > -			bad_object(offset, "inflate returned %d", ret);
> > +	do {
> >  		stream.next_in = fill(1);
> >  		stream.avail_in = input_len;
> > -	}
> > +		status = git_inflate(&stream, 0);
> > +		use(input_len - stream.avail_in);
> > +	} while (status == Z_OK);
> >  	git_inflate_end(&stream);
> > +	if (stream.total_out != size || status != Z_STREAM_END)
> > +		bad_object(offset, "inflate returned %d", status);
> >  	return buf;
> 
> Is stream.total_out still valid after inflateEnd() (and I mean "valid by
> definition", not just "by accident")? 

I don't see any indication to the contrary.

> To stay on the safe side, you should call git_inflate_end only after 
> this last check.

Agreed.  Will do.


Nicolas