From: Junio C Hamano <gitster@pobox.com>
Date: Sun, 11 Sep 2016 21:42:38 GMT
Subject: Re: [PATCH v7 06/10] pkt-line: add functions to read/write flush terminated packet streams
Message-ID: <xmqq4m5m865t.fsf@gitster.mtv.corp.google.com>
In-Reply-To: <CAGZ79kYHX6-hhKW_6JEd1hBNi8R+sBY-RjZARfHDMnyLKFZnsQ@mail.gmail.com>

Stefan Beller <sbeller@google.com> writes:

> So as a caller the strbuf is in a different state in case of error
> depending whether
> the strbuf already had some data in it. I think it would be better if
> we only did
> `strbuf_setlen(sb_out, oldlen);` here, such that the caller can
> strbuf_release it
> unconditionally.

If the caller _knows_ that the strbuf is no longer needed, it can
unconditionally call strbuf_release() on it, whether its buffer was
already released or only its length was set to 0, no?

The callee is merely trying to be nice by resetting the strbuf to a
state close to the original in the error return codepath, I would
think.  It may be debatable if such a niceness is needed, but it is
a different matter that does not relate to the burden imposed on the
caller.