From: Junio C Hamano <gitster@pobox.com>
Date: Tue, 12 Jan 2016 17:50:37 GMT
Subject: Re: [PATCH v10] ls-files: add eol diagnostics
Message-ID: <xmqqd1t6vg9e.fsf@gitster.mtv.corp.google.com>
In-Reply-To: <56949C53.5030605@web.de>

Torsten Bögershausen <tboegi@web.de> writes:

> On 01/11/2016 06:48 PM, Junio C Hamano wrote:
>
> Thanks for comments
>>> +		if (line_terminator == '\n')
>>> +			printf("i/%-6s w/%-6s attr/%-9s ", i_txt, w_txt, a_txt);
>> Can we do something better than these hard-coded constants?  Why
>> can't the "one HT between each" approach be used for both?
> v11 will make more clear, that currently "eol=crlf" is the same as
> "text eol=crlf".
> The shortest attr is then "", the longest "text eol=crlf".
> Using '\t' as a separator makes the output cluttered for a human reader.
>
> When parsing the output, the '\t' is much more convenient (and that's
> what t0027 wil use).
> Is there any chance to have a selection between human-friendly and
> script-friendly ?
> Would an option -e (eol info using spaces) vs --eol (eol info with
> TABs) make sence ?

I do not think so; what you have in v10 is infinitely better.

What I found disturbing the most was these hardcoded numbers 6,6,9
in the printf format.