From: Giuseppe Bilotta <giuseppe.bilotta@gmail.com>
Date: Fri, 03 Dec 2010 22:22:08 GMT
Subject: Re: [PATCHv2 4/7] web--browse: support opera, seamonkey and elinks
Message-ID: <AANLkTi=DdQd5MgU2bp5sb5b5t1bzXpwf8WSJ=boSbc5q@mail.gmail.com>
In-Reply-To: <7vei9yms6u.fsf@alter.siamese.dyndns.org>

On Fri, Dec 3, 2010 at 11:01 PM, Junio C Hamano <gitster@pobox.com> wrote:
> Giuseppe Bilotta <giuseppe.bilotta@gmail.com> writes:
>
>> -w3m|links|lynx|open)
>> +w3m|elinks|links|lynx|open)
>>       eval "$browser_path" "$@"
>>       ;;
>>  start)
>>       exec "$browser_path" '"web-browse"' "$@"
>>       ;;
>> -dillo)
>> +opera|dillo)
>>       "$browser_path" "$@" &
>>       ;;
>
> Not a complaint on your patch, but is there a reason we say "eval", "exec"
> and "(nothing)" in these three case arms?
>
> The above makes the interpretation of $browser_path and $@ inconsistent
> between lynx family codepath (which would apply $IFS to find the browser)
> and dillo and start family codepath (which would not), and I am wondering
> if that difference is intended.

I was wondering about this myself, but since I couldn't come up with
an explanation I opted for leaving things as they were. Lacking an
explanation I can also provide a patch to standardize invocation.

-- 
Giuseppe "Oblomov" Bilotta