From: Ramkumar Ramachandra <artagnon@gmail.com>
Date: Sun, 13 Nov 2011 10:44:12 GMT
Subject: Re: [PATCH 3/5] sequencer: sequencer state is useless without todo
Message-ID: <CALkWK0nGhUshwJM1vmAUhBG9foH+=6+_KFhfTTF6+kNS0Hm2JA@mail.gmail.com>
In-Reply-To: <20111106002645.GE27272@elie.hsd1.il.comcast.net>

Hi,

Jonathan Nieder wrote:
> Ramkumar Ramachandra wrote:
> [...]
> The usual commit-message debugging strategy applies here: imagine you
> are a BIOS clone manufacturer, and for legal reasons you are not
> allowed to read this part of the git implementation embedded in the
> standard BIOS.  However, you are allowed to read the commit message,
> and if that message is clear enough, it will explain the purpose and
> behavior of that code and you will be able to implement a compatible
> implementation addressing the same problem without scratching your
> head too much.

Ah, it helps to think about commit messages like this.  Thanks.

>> --- a/sequencer.c
>> +++ b/sequencer.c
>> @@ -654,11 +654,15 @@ static void walk_revs_populate_todo(struct replay_insn_list **todo_list,
>>
>>  static int create_seq_dir(void)
>>  {
>> +     const char *todo_file = git_path(SEQ_TODO_FILE);
>>       const char *seq_dir = git_path(SEQ_DIR);
>
> Scary idiom.

What's scary about it?

-- Ram