svn commit: r316477 - head/usr.bin/grep

[Bruce Evans]

On Tue, 4 Apr 2017, Andrey Chernov wrote:

> On 04.04.2017 2:16, Ed Maste wrote:
>>  			if (color)
>> -				fprintf(stdout, "\33[m\33[K");
>> +				fprintf(stdout, "\33[00m\33[K");
>
> Please back that one out. We don't need to handle internally or print
> remotely excessive 00.
> At least according to
> https://en.wikipedia.org/wiki/ANSI_escape_code
> "With no parameters, CSI m is treated as CSI 0 m (reset / normal), which
> is typical of most of the ANSI escape sequences."

This might be to work around a bug in a terminal driver.  I needed the
0 (but just one) to reset the attributes in my attribute tests, on at
least 1 of the tested terminals (current sc, current vt, old sc, old
xterm, old cygwin xterm).  CSI m seems to be working for sc now, and the
teken code looks OK.

I first thought that this was a misplaced 0 for \033.  2 digits for
octal looks strange.  I would use hex in C anyway.  In my user attribute
tests, I have to use printf(1), and noticed that printf(1) is missing too
many features in printf(3), or worse, is incompatible, especially for hex
escape sequences and %c format.

Bruce