From: Junio C Hamano <gitster@pobox.com>
Date: Thu, 15 Mar 2018 18:37:44 GMT
Subject: Re: [PATCH v5 01/35] pkt-line: introduce packet_read_with_status
Message-ID: <xmqq7eqd5gnb.fsf@gitster-ct.c.googlers.com>
In-Reply-To: <20180315172848.GB174336@google.com>

Brandon Williams <bmwill@google.com> writes:

>> EOF was -1 and NORMAL was 0 in the previous round; do we need to
>> read through all the invocations of functions that return this type
>> and make sure there is no "while (such_a_function())" that used to see
>> if we read NORMAL that is left un-updated?
>>  ...
>> Will replace.  Thanks.
>
> A reviewer in the previous round found that it was unnecessary to have
> EOF start at -1, so per their comments I got rid of that.

Yes, I am aware of that exchange, and after vetting the callers I
think it is "unnecessary" for EOF to be negative and NORMAL to be 0
with the current code (iow, any value can be used for these enums as
long as they are distinct).

But that is different matter.  If having negative EOF and/or zero
NORMAL helps readability of the resulting code, then even if it is
not "necessary" for EOF to be negative, it would still be "better".