[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20191225123607.38be4bdc@hermes.lan>
Date: Wed, 25 Dec 2019 12:36:07 -0800
From: Stephen Hemminger <stephen@...workplumber.org>
To: Peter Junos <petoju@...il.com>
Cc: netdev@...r.kernel.org
Subject: Re: [PATCH] ss: use compact output for undetected screen width
On Mon, 23 Dec 2019 13:47:16 +0100
Peter Junos <petoju@...il.com> wrote:
> This change fixes calculation of width in case user pipes the output.
>
> SS output output works correctly when stdout is a terminal. When one
> pipes the output, it tries to use 80 or 160 columns. That adds a
> line-break if user has terminal width of 100 chars and output is of
> the similar width.
>
> To reproduce the issue, call
> ss | less
> and see every other line empty if your screen is between 80 and 160
> columns wide.
>
> Signed-off-by: Peter Junos <petoju@...il.com>
I would prefer that if the use pipes the command output to a pipe that
the line length was assumed to be infinite.
> @@ -1159,7 +1159,13 @@ static int render_screen_width(void)
> */
> static void render_calc_width(void)
> {
> + bool compact_output = false;
> int screen_width = render_screen_width();
> + if (screen_width == -1) {
> + screen_width = 80;
> + compact_output = true;
> + }
> +
> struct column *c, *eol = columns - 1;
> int first, len = 0, linecols = 0;
>
With this patch, declarations and code are now mixed (more than before).
I would expect something like:
static void render_calc_width(void)
{
int screen_width, first, len = 0, linecols = 0;
bool compact_output = false;
struct column *c, *eol = columns - 1;
screen_width = render_screen_width();
if (screen_width == -1) {
screen_width = INT_MAX;
compact_output = true;
}
Powered by blists - more mailing lists