-------------------------------------------------------------------------------
Printf in scripting languages
-------------------------------------------------------------------------------
Type Checking

The C programming library (from which printf() derived), is very strict on the
type for the arguments given to the function (any function actually).

The Bash built-in or external command printf also has this strict type
checking, reporting an error if it is not the case.  Which is strange
considering all arguments are in reality, just strings.

  printf "%d\n" 3.9
  -bash: printf: 3.9: invalid number
  3
  /bin/printf "%d\n" 3.9
  /bin/printf: ‘3.9’: value not completely converted
  3

Perl and awk (which is typed), will silently truncate the number to become an
integer, as you would expect.

  perl -e 'printf "%d\n", 3.9;'
  3
  awk 'END{printf "%d\n", 3.9;}' /dev/null
  3

-------------------------------------------------------------------------------
Printf and UTF-8 strings

The POSIX requirements specify that the C printf() is to do field calculations
in bytes for backward compatibity.  As sucn many "printf" function will display
field shorter than expected if the string contains UTF-8 characters just as it
would if the string contains control codes for color or bold.

The library UTF-8 printf() is wprintf()

See 
  Why is printf “shrinking” umlaut?
    https://unix.stackexchange.com/questions/350240
  printf: multibyte characters
    https://unix.stackexchange.com/questions/405113/
  UTF-8 Width Display Issue of Chinese Characters (posible solutions)
    https://stackoverflow.com/questions/10751867/

Here is a list of what understands UTF-8 string lengths for field size.

  POSIX complient (not UTF-8 compatible)
  Output field is only 14 characters, rather than the requested 30!
  EG the result is only:  |✓✔√⍻✗✘✕✖      |

    Bash
      printf "|%-30s|\n" "✓✔√⍻✗✘✕✖"

    GNU Command
      /bin/printf "|%-30s|\n" "✓✔√⍻✗✘✕✖"

    perl direct
      perl -e 'printf "|%-30s|\n", "✓✔√⍻✗✘✕✖"'

  UTF-8 compatible...
  Result is: |✓✔√⍻✗✘✕✖                      |

    awk
      awk 'END{printf "|%-30s|\n", "✓✔√⍻✗✘✕✖";}' /dev/null

    perl done correctly
      perl -CO -Mutf8 -e 'printf "|%-30s|\n", "✓✔√⍻✗✘✕✖"'

    python formated string
      python -c 'print("|%-30s|" % "✓✔√⍻✗✘✕✖" )'
    multiple args
      python -c 'print("|%*s|" % (-30,"✓✔√⍻✗✘✕✖") )'

The only good solution is to do the display width to field width calculations
yourself, especially as passing random strings to other programs (like python)
is not as simple as you may think.

BASH at least gives character length in its ${#var} output.

Other things to watch out for or avoid if posible...

   * Newlines, returns, tabs, and other control codes
   * Color Escapes
   * TTY escape codes (moving cursor around)
   * TTY Title Bar Set Codes
   * Double-Wide Characters (Chinese)
   * Zero-width characters
   * Emoji
   * Proportional fonts on display output!  (game-over for TTY's)

BASH avoids many of this in it prompt handled by requireing the user to
identify such control sequences.

-------------------------------------------------------------------------------
Number Formatting (printf)

WARNING: Many programs (bash, perl, awk, printf) treat a number with a leading
'0' as being Octal. But what if it isn't a Octal, but Decimal!
This can depend if the argument is being treated as a string or as number.

  printf "%d\n" 013                            # bash builtin - BAD
  11
  /bin/printf "%d\n" 013                       # external cmd - BAD
  11

  perl -e 'printf "%d\n", 013;'                # BAD treated as octal
  11
  perl -e 'printf "%d\n", "013";'              # GOOD if input is a string
  13

  awk 'END{printf "%d\n", 013;}' /dev/null     # BAD treated as octal
  11
  awk 'END{printf "%d\n", "013";}' /dev/null   # GOOD if input is a string
  13

To fix in BASH use the expression $((10#$num)), which will ensure the number
is treated as decimal, and removes any leading zeros from the number, before
giving it to the printf command (built-in or otherwise).

  echo $((10#013))
  13
  printf "%d\n" $((10#013))
  13

-------------------------------------------------------------------------------
Number Decimal Places...

Using %.2g  will take a floating point number and only output 2 significant
digits but keep the result as short as posible!

  printf "%.2g\n" 3.38    # -> "3.4"
  printf "%.2g\n" 4.01    # -> "4"

contrast this with the more normal "%.2f" formating...

  printf "%.2f\n" 3.377   # -> "3.38"
  printf "%.2f\n" 4       # -> "4.00"

-------------------------------------------------------------------------------
BASH Printf Date functions

BASH printf has a  '%(date_format)T' formatting,
taking an argument of a date in seconds.

If argument is  -1   use the time NOW,  -2  time when BASH started

As it is a builtin, the output can be saved directly into a BASH variable!

   # Save current time in seconds into variable!
   printf -v now '%(%s)T' -1

   # print saved time in IEEE date format
   printf '%(%F_%T)T' $now



Example...
Progressive Time out waiting for condition....
With fancy rotating line...

  # Initialization
  printf -v start '%(%s)T' -1  # start time
  heartbeat='-\|/';
  beatlen=4
  timeout=30

  # Wait loop
  while true; do
    # wait for a server to appear -- adjust to suit
    # http_code=$(curl -s -o/dev/null -w'%{http_code}' https://localhost:8443/)
    # case "$http_code" in
    #   501|504) : ;;  # not started yet
    #   302) break ;;  # SUCCESS - break loop
    #   *)   echo "Unknown HTTP response $http_code"; break ;;
    # esac
    printf -v elapsed '%(%s)T' -1  # elapsed time
    if (( elapsed - start > timeout )); then
      printf "FAILED"
      break   #exit 10
    fi
    beat=$(( (beat += 1)%beatlen ))
    printf "%c Waiting... %d\r" "${heartbeat:$beat:1}" $(( elapsed - start ))
    sleep .1    # max speed of rotating line
  done
  printf "                        \n"

-------------------------------------------------------------------------------