# 24 and 12 Hour Times

Write a program or function with no input that prints or returns this string of 24-hour and 12-hour times:

00:00 12:00am
01:00  1:00am
02:00  2:00am
03:00  3:00am
04:00  4:00am
05:00  5:00am
06:00  6:00am
07:00  7:00am
08:00  8:00am
09:00  9:00am
10:00 10:00am
11:00 11:00am
12:00 12:00pm
13:00  1:00pm
14:00  2:00pm
15:00  3:00pm
16:00  4:00pm
17:00  5:00pm
18:00  6:00pm
19:00  7:00pm
20:00  8:00pm
21:00  9:00pm
22:00 10:00pm
23:00 11:00pm


The string must be output exactly as it appears here. The only exception is that it may optionally have a single trailing newline.

So the MD5 hash of your output should be

827ae6e2dbb1df494930baedb3ee2653


if you do not have a trailing newline and

cd4c3d18abee9bafb495f390a919a13f


if you do. (Your hash could be different if your system uses another type of newline but that's OK.)

The shortest code in bytes wins. Tiebreaker is earlier answer.

• – Joshpbarron May 31 '16 at 8:01

# Bash + coreutils, 43 30

• Saved 7 bytes thanks to @Yossarian
• Saved 3 bytes thanks to @AndersKaseorg
seq 0 23|date -f- +%R\ %l:00%P

• seq generates integers 0-23, one per line.
• date interprets each line as a datetime. Bare integers appear to be sufficient to be recognised as hours of the day by date. date then outputs each time with the required formatting using the available time format specifiers.

Assumes LANG=C locale, as per this meta answer.

Ideone.

• You can even just do seq 0 23 and save 7 bytes – Yossarian Jun 1 '16 at 8:21
• And %H:00%R saves another 3. – Anders Kaseorg Jun 1 '16 at 11:07
• @AndersKaseorg, nop, %R inserts a leading zero, not wanted by the OP. – rexkogitans Jun 1 '16 at 11:44
• @rexkogitans A leading zero is wanted (in the first column, which is the one I’m talking about). – Anders Kaseorg Jun 1 '16 at 11:54
• @AndersKaseorg, oh, you were talking about the first one... then you are right, of course. Corrected my comment above. – rexkogitans Jun 1 '16 at 12:27

# Python 2, 66 bytes

for i in range(24):print'%02d:00%3d:00%sm'%(i,12+i%-12,'ap'[i>11])

• modulo -12 is clever! – Erik the Outgolfer May 31 '16 at 8:00
• Just in case anyone is wondering: i=0;exec"print'%02d:00%3d:00%cm'%(i,~-i%12+1,97+i/12*15);i+=1;"*24 changes the code in three different ways, but each change is the same length. – Sp3000 May 31 '16 at 8:31

# C, 73 bytes

m(i){for(i=25;--i;)printf("%02d:00%3d:00%cm\n",24-i,12-i%12,"pa"[i/13]);}


mIllIbyte found a particularly neat way to rewrite this answer. Thanks!

• Cool - I never knew about ?: – Digital Trauma May 31 '16 at 17:27
• I didn’t know! Interesting. I fixed it :) – Lynn May 31 '16 at 18:47
• Ahh, well, ?: is actually the closest equivalent to Perl’s ||, I believe. C’s || is more like (a || b) ? 1 : 0. – Lynn May 31 '16 at 20:48
• i++,i%12?:12,"ap"[i/12]) lacks a sequence point. Can't be certain when i++ occurs. Perhaps for(i=0;i<24;i++) ...i,i%12?:12,"ap"[i/12]... – chux - Reinstate Monica Jun 1 '16 at 4:26
• i++<24 is better - doesn't add more bytes to the source code – anatolyg Jun 1 '16 at 7:17

# MATL, 4642 34 bytes

12tEt:qy/t15XObZ"!b16XOhhkw14:X~Z)


Previously, 42 bytes, 12tEt:q2M/736330+t15XObZ"!b16XOhhkw14:X~Z), and 46 bytes, 736330 24t:qw/+t15XO' '24TX"b16XOhhk14: 12X~Z). Of course the 736330 wasn't needed, that was crazy!

Note: Doesn't work with TryItOnline, I think there is a compatibility issue between Matlab and Octaves implementation of datestr.

datestr takes the number representation of a date and converts it to the string representation of that date. The time of the day is the fractional part of the number, so 0.0 corrsponds to January 0, 0000, at time 00:00:00, and 1.0 corresponds to January 1, 0000, at 00:00:00. 1/24 is 1am, 2/24 2am etc.

Explanation

12t         % push a 12 onto the stack and duplicate
Et          % double the 12 and duplicate the 24 (stack now has 12, 24, 24, bottom to top)
:q          % make vector 1:24 and decrement by 1, stack has 12, 24, 0:23
y           % duplicate second element on stack (24)
/           % divide, for (0:23)/24
t           % duplicate elements
15XO        % string representation of date, 15 specifies format
b           % bubble up element in stack (gets a 24 on top of the stack)
Z"!         % makes a column of 24 spaces, to put between columns of times
b           % bubble up another (0:23)/24
16XO        % string representation of date, 16 for a different format
hh          % concatenate two time vectors and the column of spaces
k           % convert string to lowercase, because CO gives AM/PM not am/pm
w           % swap elements in stack, that first 12 is now on top
14:         % vector of equally spaced values 1:14
X~          % set exclusive-or, returns [1 2 3 4 5 6 7 8 9 10 11 13 14]
Z)          % get the right columns of the string array to remove extra column of blanks
% implicit display


To show it works in Matlab, here is a screenshot

# ///, 160 bytes

/Z/:00 //S/Z //A/:00am
//P/:00pm
/00Z12A01S1A02S2A03S3A04S4A05S5A06S6A07S7A08S8A09S9A10Z10A11Z11A12Z12P13S1P14S2P15S3P16S4P17S5P18S6P19S7P20S8P21S9P22Z10P23Z11P


Try it online!

Ungolfed

00:00 12:00am
01:00  1:00am
02:00  2:00am
03:00  3:00am
04:00  4:00am
05:00  5:00am
06:00  6:00am
07:00  7:00am
08:00  8:00am
09:00  9:00am
10:00 10:00am
11:00 11:00am
12:00 12:00pm
13:00  1:00pm
14:00  2:00pm
15:00  3:00pm
16:00  4:00pm
17:00  5:00pm
18:00  6:00pm
19:00  7:00pm
20:00  8:00pm
21:00  9:00pm
22:00 10:00pm
23:00 11:00pm


• Fun fact: making the :00 into a replacement is one byte longer... TIO – steenbergh Feb 1 '17 at 9:42
• @steenbergh That's because :00 is a 3-byte string that appears 3 times. 3 × 3 = 9, making a replacement in your case costs 3+1+3=7 since it uses a 1-byte alias, and you use it 3 times, so 7+3=10. 9 < 10, so I won't replace it. – Erik the Outgolfer Feb 1 '17 at 11:49

# MarioLANG, 965 834 bytes

Try it online

well this was ridiculously complicated.

Technically the output is valid but in practice the Tio for MarioLANG output "n " instead of "n" when we print a number with ':'

if I find the time i'll guess i'll try to do a (probably much longer) version of the program that output correctly on Tio

++<>) +++@++++>   [!) >)>((((::(.)::((.))+:+:--(.)::)).).).)+++++++++
++""+ +"=====""====#) "+"============================================
+++)+ +>>+++++- <+<)->+++ ![-).).).))(::)).(:(:)))..(((::)).(:+(:((((<
+>+++ ++"====<( ")")-"!+++#=========================================="
+(+++>++!++)<+( ++++-+++++>
-))+)=(#==="+( ++++)+++++"==========================================!
[!!+-[!(+++!!! !+!<+!++!>(((((+:(.))::(((.
==##===#====###=#=#"=##=#"=================<
++++)))+++++++++++++++(((![-).).).)::)).(:))   >
>========================#==================   "
+>+ >
+"+ "=======================================[   =====================#===============[
+!> ! -).).).))(::)).)):+(..(((::)).(:+(((((<++!-).).).))(::)).)):+(.(((::)).(:+(((((<
=#==#======================================="==#======================================


Explanation :

our main problem here is the fact that we have 6 NaN char (newLine, Space, :, a, p, m)

in marioLANG, in order to print character, we need their ascii value:

• newLine is 10
• Space is 32
• : is 58
• a is 97
• p is 112
• m is 109

So the first thing to do is to set the memory :

++<>) +++@++++>   [!) >)> !
++""+ +"=====""====#) "+"==
+++)+ +>>+++++- <+<)->+++
+>+++ ++"====<( ")")-"!+++
+(+++>++!++)<+( ++++-+++++
-))+)=(#==="+( ++++)+++++
[!!+-[!(+++!!! !+!<+!++!>
==##===#====###=#=#"=##=#"


with this, the memory look like :

                   v
32 58 0 0 97 109 10 0
_  :      a   m  \n


we will transform a into p during the rest of the program

then we do the actual output :

++<>) +++@++++>   [!) >)>((((::(.)::((.))+:+:--(.)::)).).).)+++++++++
++""+ +"=====""====#) "+"============================================
+++)+ +>>+++++- <+<)->+++ ![-).).).))(::)).(:(:)))..(((::)).(:+(:((((<
+>+++ ++"====<( ")")-"!+++#=========================================="
+(+++>++!++)<+( ++++-+++++>
-))+)=(#==="+( ++++)+++++"==========================================!
[!!+-[!(+++!!! !+!<+!++!>(((((+:(.))::(((.
==##===#====###=#=#"=##=#"=================<
++++)))+++++++++++++++(((![-).).).)::)).(:))   >
>========================#==================   "
+>+ >
+"+ "=======================================[   =====================#===============[
+!> ! -).).).))(::)).)):+(..(((::)).(:+(((((<++!-).).).))(::)).)):+(.(((::)).(:+(((((<
=#==#======================================="==#======================================


# Julia, 887166 64 bytes

[@printf "%02d:00%3d:00%cm
" i-11 i%12+1 i>22?112:97for i=11:34]


This is a full program that prints the string with a single trailing newline.

Try it online!

Saved 5 bytes thanks to Sp3000 and 2 thanks to Dennis!

## C# function, 100 bytes

void F(){for(int i=0;i<24;i++){Console.Write($"{i:00}:00 {(i+11)%12+1,2}:00 {(i>11?"p":"a")}m\n");}}  Ungolfed version: void F() { for (int i = 0; i < 24; i++) { Console.Write($"{i:00}:00 {(i + 11)%12 + 1,2}:00 {(i > 11 ? "p" : "a")}m\n");
}
}


Console.Write() takes too many characters!

• Hello, and welcome to PPCG! I use C# a bit, but what is the leading $ doing to the string? – NoOneIsHere Jun 1 '16 at 1:48 • @NoOneIsHere - That's called "string interpolation" and it's a new feature in C# 6.0. You can read more about it here https://msdn.microsoft.com/en-us/library/dn961160.aspx. – STLDev Jun 1 '16 at 3:15 ## JavaScript (ES2015), 147138137134 133 bytes ((o,x,r)=>{for(i=0;i<24;)b=i%12,c=b||12,o+='0'[r](i<10)+i+++x+' '[r]((c<10)+1)+c+x+(i<13?'a':'p')+"m\n";return o})('',':00','repeat')  In this version I took advantage of String.repeat() method to get rid of lengthy .slice() and .join() and moved incrementation inside loop. Previous version: ((o,x,i)=>{for(;i<24;i++){b=i%12;o+=[0${i+x}.slice(-5),(b||12)+x+(i<12?'a':'p')+'m'].join(' '.repeat((b>0&&b<10)+1))+"\n"}return o})('',':00',0)


Gives output with trailing newline. Tested in Firefox Scratchpad. Not sure if passing arguments to IIFE is OK with "no input" rule.

It's my first submission, so hello to everyone! :)

# TSQL(SQLServer 2012) 146124 121

DECLARE @ DATETIME=0WHILE @<1BEGIN PRINT


Try it online!

First attempt, a bit longer, but a one-liner:

SELECT CONVERT(char(5),n,108)+' '+LOWER(RIGHT(FORMAT(n,'g'),8))FROM(SELECT
top 24 dateadd(hh,Number,0)n FROM master..spt_values WHERE'P'=type)x


Try it online!

• Reduced to 121 chars: try it online – Ross Presser Jun 14 '16 at 14:58
• @RossPresser once again you helped me, learning alot here – t-clausen.dk Jun 14 '16 at 15:06

# Perl 5, 58

map{printf"%02u:00%3u:00%sm
",$_,$_%12||12,$_>11?p:a}0..23  # Javascript, 122 Bytes, 120 Bytes f=j=>j>12?j-12:j;for(i=0;i<24;i++)console.log('%s:00 %s:00%s',i<10?'0'+i:i,i==0?12:f(i)<10?' '+f(i):f(i),i>11?'pm':'am')  Edit: Small bug fixed + output: 00:00 12:00am 01:00 1:00am 02:00 2:00am 03:00 3:00am 04:00 4:00am 05:00 5:00am 06:00 6:00am 07:00 7:00am 08:00 8:00am 09:00 9:00am 10:00 10:00am 11:00 11:00am 12:00 12:00pm 13:00 1:00pm 14:00 2:00pm 15:00 3:00pm 16:00 4:00pm 17:00 5:00pm 18:00 6:00pm 19:00 7:00pm 20:00 8:00pm 21:00 9:00pm 22:00 10:00pm 23:00 11:00pm  # V, 56 53 bytes i00:00 23ñYpñH12G$yP13G$pgvó$/am
í/pm
í 0/
í/12


Try it online!

Since this can be hard to enter, here is a reversible hexdump:

00000000: 6930 303a 3030 201b 3233 f159 7001 f148  i00:00 .23.Yp..H
00000010: 1631 3247 2479 5031 3347 2470 6776 f320  .12G$yP13G$pgv.
00000020: 242f 616d 0aed 2f70 6d0a ed20 302f 2020  $/am../pm.. 0/ 00000030: 0aed 2f31 320a ../12.  A noncompeting version is trivially 2 bytes shorter if you replace both occurrences of G$ with L, which was supposed to be the same but had a bug.

Explanation:

i00:00<esc>                                     #Enter the starting text.
23ñYp<C-a>ñ                          #Duplicate and increment 23 times
H                         #Move back to the beginning
<C-v>12G$y #Select 12 lines horizontally P #Horizontally paste 13G$p         #Move to line 13 and Horizontally paste again
gv       #Reselect the top 12 lines
ó $/am #Replace a space at the end of the line with 'am'  í/pm #Replace the previous search with 'pm' í 0/ #Replace "Space+0" with 2 spaces í/12 #Replace the previous search with "12"  ## 05AB1E, 51504844 42 bytes Saved two bytes thanks to carusocomputing Code: 24FNgi0}N…:00©ðN12(%12+Dgiðs}®„paN12‹è'mJ,  Try it online! Explanation 24F # for N in [0...23] Ngi0} # if len(N)=1, push 0 N # push N …:00© # push ":00" and store a copy in register ð # push " " N12(%12+D # push 2 copies of N%(-12)+12 giðs} # if the length of that number is 1, # push " " and swap with the number ® # push ":00" again „pa # push "pa" N12‹è # index into that with N<12 'm # push "m" J, # join everything and print with newline  • codegolf.stackexchange.com/questions/103242/… ;). Now whether or not that could be optimized below 50 is up for question haha. – Magic Octopus Urn Jan 31 '17 at 21:14 • 23Ý instead of 24L< for 1 byte. And how long has ë existed? I feel so dumb not knowing about else statements in 05AB1E until now. – Magic Octopus Urn Jan 31 '17 at 21:18 • @carusocomputing: Thanks! else has existed for a while, but it's been buggy at times. Especially when nesting ifs. – Emigna Jan 31 '17 at 21:22 ## PowerShell v2+, 76 bytes 0..23|%{"{0:D2}:00{1,3}:00"-f$_,(($_%12),12)[!($_%12)]+('am','pm')[$_-ge12]}  Loops from 0..23 and each loop sets a string with the -f operator. The first {0:D2} ensures we have prepended zeros, the second {1,3} ensures we have padded spaces for the middle column. The {0} one corresponds to the $_ of the -f operator, while the {1} corresponds to the pseudo-ternary that chooses between $_%12 or 12 based on whether $_%12 is non-zero or not (i.e., if we're at $_=13, this will choose 1 for 1:00pm). We then concatenate that with another pseudo-ternary that chooses the appropriate am/pm. As opposed to my answer on List all times in the day at a half hour rate, it's actually shorter here to brute-force the numbers since we get significantly cheaper padding. Here's the answer using date functions, at 78 bytes 0..23|%{(Date -h$_ -f 'HH:00')+(Date -h $_ -f "h:00tt").ToLower().PadLeft(8)}  # C++, 81 79 bytes []{for(time_t t=0,y;t<24;cout<<put_time(gmtime(&y),"%R %l:00%P\n"))y=t++*3600;}  This code requires using namespace std somewhere preceding it. It does a loop on the values 0...23. It multiplies each value by 3600, converts to a tm struct and prints it. The print format %R outputs the 24-hour and minute; the print formats %l and %P output the proper 12-hour parts; they require GNU. A working online version is here. # Ruby, 66 62 bytes 0.upto(23){|i| puts "%02d:00%3d:00#{'ap'[i/12]}m"%[i,(i-1)%12+1]}  New version 24.times{|i|puts"%02d:00%3d:00#{'ap'[i/12]}m"%[i,(i-1)%12+1]}  • 24.times is shorter. No need for the spaces around puts. – manatwork Jun 1 '16 at 12:05 • You can replace (i-1) with ~-i for 2 bytes. – Jordan Nov 7 '17 at 18:55 # JavaScript (ES6), 119 116 bytes _=>Array(24).fill().map((_,i)=>${0${i}.slice(-2)}:00${ ${(i+11)%12+1}.slice(-2)}:00${'ap'[+(i>11)]}m).join



# Sclipting, 76 bytes

The program assumes that the input is empty (or '0' or anything that converts to the integer 0).

The byte count assumes UTF-16 encoding.

군上❶겠小꼀虛嗎❷꾣갰글❷결加곀剩增❶겠小글虛嗎댆밁⓷꾣갰⓷⓼곀小掘닐밊終

Ungolfed:

군 // 23

❶겠小꼀虛嗎 // n < 10 ? "0" : ""
❷          // n
꾣갰글      // ":00 "
❷결加곀剩增 // k = (n+11) % 12 + 1
❶겠小글虛嗎 // k < 10 ? " " : ""
댆밁       // "pa"
⓷         // Pull n to top of stack
꾣갰       // ":00"
⓷         // Pull "pa" to top of stack
⓼         // Pull k to top of stack
곀小掘     // "pa"[k < 10 ? 1 : 0]
닐밊       // "m\n"



Each iteration of the loop leaves lots of small strings on the stack; at the end they are all automatically concatenated.

# JavaScript, 97 95 bytes

This is based off of starcorder’s answer. Thanks to George Reith for a 2 byte improvement.

for(i=0,k=12;i<24;k=i++%12+1)console.log('%s:00 %s:00%sm',i>9?i:'0'+i,k>9?k:' '+k,i>11?'p':'a')


Ungolfed:

for (i=0, k=12; i < 24; k = (i++) % 12 + 1)
console.log('%s:00 %s:00%sm',
i > 9 ? i : '0' + i,
k > 9 ? k : ' ' + k,
i > 11 ? 'p' : 'a')

• You can save 2 bytes by rewriting n < 10 as 9 < n and swapping the ternary cases – George Reith Jul 13 '16 at 11:40

## Batch, 167 bytes

@echo off
set h=11
set p=a
for /l %%a in (0,1,23)do call:e %%a
exit/b
:e
set a=0%1
set/ah=h%%12+1
set h= %h%
if %1==12 set p=p
echo %a:~-2:00 %h:~-2%:00%p%m


# Kotlin, 95 bytes

It can be improved for sure.

fun p(){for(i in 0..23)println("%02d:00 ${(i+11)%12+1}:00${if(i>12)"p" else "a"}m".format(i))}


# PHP, 110 107 bytes

for($h=0;$h<24;){$m=($h+11)%12+1;echo($h<10?0:"")."$h:00 ".($m<10?" ":"")."$m:00".($h++<12?"a":"p")."m\n";}  exploded view for ($h=0; $h<24; ) {$m = ($h+11) % 12 + 1; echo ($h < 10 ?  0  : "") . "$h:00 " . ($m < 10 ? " " : "") . "$m:00" . ($h++ < 12 ? "a" : "p") . "m\n";
}


Somewhat surprised, tried to turn the ($i < 10 ?$s : "") . "$i:00" bit into a function, but it wound up adding ~25 characters. No go there. ## Swift, 85 bytes for x in 0...23{print(String(format:"%02d:00 %2d:00\(x<12 ?"a":"p")m",x,12+x % -12))}  • This doesn't work, at least not for me on Swift 2.2. Output – JAL Jun 1 '16 at 22:14 • I had to change your format string: String(format: "%02d:00 %2d:00\(x<12 ?"a":"p")m", x, x%12 != 0 ? x%12 : 12) – JAL Jun 1 '16 at 22:15 # C Function, 82 bytes m(i){for(;i<24;printf("%02d:00 %2d:00%cm\n",i,i%12==0?12:i%12,i>11?'p':'a'),i++);}  Usage, 94 Byte m(i){for(;i<24;printf("%02d:00 %2d:00%cm\n",i,i%12==0?12:i%12,i>11?'p':'a'),i++);}main(){m();}  Ungolfed, 337 Bytes #include <stdio.h> void m(){ int i,a; char c; for(i=0;i<24;i++){ if (i%12==0){ a = 12; } else{ a = i%12; } if (i>11){ c = 'p'; } else{ c = 'a'; } printf("%02d:00 %2d:00%cm\n",i,a,c); } } int main(){ m(); }  it works on Windows: # C Program, 85 bytes main(i){for(;i<24;printf("%02d:00 %2d:00%cm\n",i,i%12==0?12:i%12,i>11?'p':'a'),i++);}  • You might need initialization i=0 for the function version. And for the program version, you can stuff the code directly into main - no need to define a function there! – anatolyg May 31 '16 at 16:44 • Thanks! edited!! FYI: In the funcion version if you use it without params it's self initialized to 0! – Giacomo Garabello May 31 '16 at 16:53 • i%12==0?12:i%12 --> i%12?i%12:12 – chux - Reinstate Monica Jun 1 '16 at 11:54 • "In the funcion version if you use it without params it's self initialized to 0!" This does not appear to be standard C. Any reference to support this? – chux - Reinstate Monica Jun 1 '16 at 11:57 • i have noticed it works only on the gcc i have installed on my windows system.. a friend of mine with linux told me it doesn't work on his pc but i don't know how to fix it for linux... – Giacomo Garabello Jun 1 '16 at 12:51 # Foo, 163 bytes Pretty brute-force approach; nothing clever here (I tried in a couple spots but it ended up being shorter not to), just wanted to give Foo a shot. Foo automatically prints anything within quotes. $c10 prints a line break. (## ... ) loops until the current cell equals ##.

"00:00 12:00am"$c10+1(10"0"$i":00  "$i":00am"$c10+1)(12$i":00 "$i":00am"$c10+1)"12:00 12:00pm"$c10+1(22$i":00 ">+1$i<":00pm"$c10+1)(24$i":00 ">+1$i<":00pm"$c10+1)


Ungolfed a bit:

"00:00 12:00am"$c10+1 (10"0"$i":00  "$i":00am"$c10+1)
(12$i":00 "$i":00am"$c10+1) "12:00 12:00pm"$c10+1
(22$i":00 ">+1$i<":00pm"$c10+1) (24$i":00 ">+1$i<":00pm"$c10+1)


Try it online

# Javascript (using external library - Enumerable) (107 bytes)

_.Range(0,24).WriteLine(x=>((x<10?"0"+x:x)+":00 "+(((h=((x+11)%12)+1))<10?" "+h:h)+":00"+(x<12?"am":"pm")))


FOR H=0TO 23?FORMAT$(%02D:00 %2D:00%Sm",H,(H+11)MOD 12+1,"ap"[H>11])NEXT  Someone found a better 24->12 hour formula than my old one, which saves 3 bytes, and 5 bytes in another program # PHP, 6765 64 bytes This uses IBM-850 encoding. for(;$i<24;)printf(~┌¤═ø┼¤¤┌╠ø┼¤¤┌îÆ§,$i,$i%12?:12,$i++>11?p:a);  With the unencoded string (66 bytes): for(;$i<24;)printf("%02d:00%3d:00%sm\n",$i,$i%12?:12,$i++>11?p:a);  Run like this: php -n -r 'for(;$i<24;)printf(~┌¤═ø┼¤¤┌╠ø┼¤¤┌îÆ§,$i,$i%12?:12,$i++>11?p:a);'  ## Tweaks • Saved 2 bytes by improving sprintf format • Saved a byte by getting rid of unnecessary space (thx @Titus) • You can use -n instead of -d error_reporting=30709. There should be no space before am/pm. Save one byte on the unencoded version with a physical linebreak. – Titus Feb 1 '17 at 20:33 # tcl, 93 set i 0;time {puts [format %02d:00%3d:00[expr$i<12?"a":"p"]m $i [expr$i%-12+12]];incr i} 24