Baselines

So the Coding Horror thread on boards.ie’s Development forum threw up an interesting post:

Originally Posted by COYW

[..] I was told not to comment my code, as it was a “waste of time”. Apparently, that is classic Agile!

A good code doesn’t need comments; a bad code doesn’t need comments either – it needs to be fixed! 

The responses were about as you’d expect, but in the course of answering, I made the following suggestion:

you still have your code from college exercises, right? No? You don’t keep a private archive of old code you wrote? So… how do you tell if you’re improving or getting worse without a baseline?

So I figured I’d take a peek again (last time I looked must have been before Calum was born) and compare it to what I’m writing today.

The really low-level stuff came off pretty well. Some code for programming a GAL chip for 3D2 (where you build a thin client from a 68008 chip and a handful of other chips and what feels like a mile of hand-wrapped wire):

[cc escaped=”true” lang=”asm” lines=”20″]
‘ GAL1 PLDASM
‘ Address Decoding

‘ 3D3 68008 Project

‘ Group 5 :
‘ Mark Dennehy 93369425
‘ Ellen Delaney

‘ This is the address decoding code for the board. It uses one
‘ GAL exclusively and the smallest selectable memory block is
‘ 1024 bytes. The CLK pin is attached to /AS via an inverter.

‘ INPUTS :
‘ A10..A19
‘ BOOT
‘ /AS
‘ OUTPUTS:
‘ EPROM1
‘ RAMROM1
‘ RAM1
‘ RAM2
‘ RAM3
‘ ACIA0
‘ ACIA1
‘ LCD
‘ SYSTEMBYTE
‘ TIMER

device 22v10

‘ Input pin definitions
as = 1
a19 = 2
a18 = 3
a17 = 4
a16 = 5
a15 = 6
a14 = 7
a13 = 8
a12 = 9
a11 = 10
a10 = 11
boot = 13

‘ Power & ground pin definitions
gnd = 12
vcc = 24

‘ Output pin definitions
/eprom1 = 14
/ramrom1 = 15
/ram1 = 16
/ram2 = 17
/acia0 = 18
/acia1 = 19
/lcd = 20
/sysbyte = 21
/timer = 22
/ram3 = 23

macro io a19*a18*a17*a16;
macro rom /a19*/a18*/a17*/a16;

start

‘ EPROM selection : $00000 -> $01FFF and boot
/eprom1 /= /as*/boot*&rom*/a15*/a14*/a13;

‘ RAM_ROM selection : $00000 -> $01FFF and not-boot
‘ $20000 -> $21FFF and boot
/ramrom1 /= /as*boot*&rom*/a15*/a14*/a13 + /as*/boot*/a19*/a18*a17*/a16*/a15*/a14*/a13;

‘ RAM 1 selection : $10000 -> $107FF
/ram1 /= /as*/a19*/a18*/a17*a16*/a15*/a14*/a13*/a12*/a11;

‘ RAM 2 selection : $10800 -> $10FFF
/ram2 /= /as*/a19*/a18*/a17*a16*/a15*/a14*/a13*/a12*a11;

‘ RAM 3 selection : $11000 -> $12FFF
/ram3 /= /as*/a19*/a18*/a17*a16*/a15*/a14*/a13*a12 + /a19*/a18*/a17*a16*/a15*/a14*a13*/a12;

‘ ACIA 0 selection : $F0000
/acia0 /= /as*&io*/a15*/a14*/a13*/a12*/a11*/a10;

‘ ACIA 1 selection : $F0400
/acia1 /= /as*&io*/a15*/a14*/a13*/a12*/a11*a10;

‘ LCD selection : $FE000
/lcd /= /as*&io*a15*a14*a13*/a12*/a11*/a10;

‘ SYSTEM_BYTE selection : $FF000
/sysbyte /= /as*&io*a15*a14*a13*a12*/a11*/a10;

‘ TIMER selection : $F1000
/timer /= /as*&io*/a15*/a14*/a13*a12*/a11*/a10;

end

[/cc]
And the assembler was about readable in most cases:

[cc escaped=”true” lang=”asm” lines=”20″]
;————————————————————
;
; Demonstration PICRAT program.
;
;————————————————————
; REVISION HISTORY :
;
; 13/10/97 First draft, preliminary modularisation
; of code.
;
;————————————————————

ERRORLEVEL 0
PROCESSOR PIC16C74A
LIST b=4
__CONFIG _BODEN_OFF & _CP_OFF & _PWRTE_OFF & _WDT_OFF & _XT_OSC
TITLE “Demonstration PICRAT program”
SUBTITLE “Version 1.00”

include

;————————————————————
; Context Saving registers

CONTEXT UDATA 0x20

int_w RES 1
int_status RES 1
int_pclath RES 1
int_fsr RES 1

CONTEXT2 UDATA 0xa0

int_w2 RES 1 ;Dummy Register

;————————————————————
; Variables used in Main Loop
Main UDATA

tmpChar RES 1

;————————————————————
;
; NAME : RESET_VECTOR
;
; FUNCTION : Executes reset interrupt service routine
;
;————————————————————

RESET_VECTOR CODE 0000
RESET_VECTOR
GLOBAL RESET_VECTOR

PAGESEL reset_isr
goto reset_isr

;——————–ROUTINE SPECIFICATION——————-
;
; NAME : INTERRUPT_VECTOR
;
; FUNCTION : Contect saving, correct ISR selection
;
; NOTES : Saves W, STATUS, PCLATH as per ex.14-1
; in datasheet
;
;————————————————————
; REVISION HISTORY :
;
;————————————————————

INTERRUPT_VECTOR CODE 0004
INTERRUPT_VECTOR
GLOBAL INTERRUPT_VECTOR

EXTERN USART_Tx_isr
EXTERN USART_Rx_isr

;INTERRUPT_VECTOR
;
; Save the W,STATUS,PCLATH and FSR registers,

movwf int_w
swapf STATUS,W
clrf STATUS
movwf int_status
movf PCLATH,W
movwf int_pclath
movf FSR,W
movwf int_fsr

; Check to see what caused the interrupt,
; Byte received ?

BANKSEL PIR1
PAGESEL USART_Rx_isr
btfsc PIR1,RCIF

; Jump to USART Rx ISR

call USART_Rx_isr

; Ready to transmit byte ?

BANKSEL PIR1
PAGESEL USART_Tx_isr
btfsc PIR1,TXIF

; Jump to USART Tx ISR

call USART_Tx_isr

; Unknown interrupt ?
; Jump to exception handler

; PAGESEL Exception
; call Exception

; Restore registers and return from interrupt.

clrf STATUS
movf int_fsr,W
movwf FSR
movf int_pclath,W
movwf PCLATH
swapf int_status,W
movwf STATUS
swapf int_w,F
swapf int_w,W

retfie

;——————–ROUTINE SPECIFICATION——————-
;
; NAME : Exception
;
; FUNCTION : Called when an unhandled interrupt
; condition occours.
;
; NOTES :
;
;————————————————————
; REVISION HISTORY :
;
;————————————————————

;EXCEPTION
; Endless loop

Exception CODE
Exception

goto Exception

;——————–ROUTINE SPECIFICATION——————-
;
; NAME : reset_isr
;
; FUNCTION : Reset Interrupt service routine
; Determines correct action to perform
; on startup.
;
; NOTES :
;
;————————————————————
; REVISION HISTORY :
;
;————————————————————

reset_isr CODE
reset_isr
GLOBAL reset_isr

EXTERN MemoryTest
EXTERN USART_init
EXTERN USART_puts
EXTERN USART_putc
EXTERN USART_hi_msg_tmp
EXTERN USART_lo_msg_tmp
EXTERN Startup_screen
; EXTERN LCD_Initialise
; EXTERN LCD_PutChar
EXTERN USART_getc

PAGESEL MemoryTest
call MemoryTest

PAGESEL USART_init
call USART_init

; PAGESEL LCD_Initialise
; call LCD_Initialise

PAGESEL USART_putc
movlw A’.’
call USART_putc
movlw A’.’
call USART_putc
movlw A’.’
call USART_putc
movlw A’\r’
call USART_putc
movlw A’\n’
call USART_putc

; PAGESEL LCD_PutChar
; movlw A’T’
; call LCD_PutChar
; movlw A’e’
; call LCD_PutChar
; movlw A’s’
; call LCD_PutChar
; movlw A’t’
; call LCD_PutChar

; Enable perihiperal interrupts

BANKSEL INTCON
bsf INTCON,PEIE

; Enable all interrupts
bsf INTCON,GIE

; Print out startup message
PAGESEL USART_puts
movlw high Startup_screen
movwf USART_hi_msg_tmp
movlw low Startup_screen
movwf USART_lo_msg_tmp
call USART_puts

bcf OPTION_REG,7
BANKSEL PORTB
clrf PORTB
BANKSEL TRISB
movlw 0x00
movwf TRISB

BANKSEL PORTD
clrf PORTD
BANKSEL TRISD
movlw 0x7F
movwf TRISD

BANKSEL PORTE
clrf PORTE
BANKSEL TRISE
movlw 0x07
movwf TRISE

PAGESEL MainLoop
call MainLoop

;——————–ROUTINE SPECIFICATION——————-
;
; NAME : MainLoop
;
; FUNCTION : Main Control Interpreter
;
; NOTES : A Finite State Machine
;
;————————————————————
; REVISION HISTORY :
; 9/1/98 First Draft
;————————————————————

MainLoop CODE
MainLoop
GLOBAL MainLoop
EXTERN USART_getc
EXTERN USART_putc
EXTERN AnalogRoot
ExTERN DigitalRoot
EXTERN CounterRoot
EXTERN PWMRoot
EXTERN MotorControlRoot
EXTERN TimerRoot

PAGESEL USART_getc
call USART_getc
BANKSEL tmpChar
movwf tmpChar
PAGESEL MainLoop
movf tmpChar,W
xorlw A’.’
btfss STATUS,Z
goto MainLoop

;————————————————————

ServiceSelect
PAGESEL USART_putc
movlw ‘-‘
call USART_putc
movlw ‘-‘
call USART_putc
movlw ‘\r’
call USART_putc
movlw ‘\n’
call USART_putc

PAGESEL USART_getc
call USART_getc
movwf tmpChar

PAGESEL AnalogRoot
movf tmpChar,W
xorlw A’a’
btfsc STATUS,Z
goto AnalogRoot

PAGESEL DigitalRoot
movf tmpChar,W
xorlw A’d’
btfsc STATUS,Z
goto DigitalRoot

PAGESEL DigitalRoot
movf tmpChar,W
xorlw A’c’
btfsc STATUS,Z
goto CounterRoot

PAGESEL PWMRoot
movf tmpChar,W
xorlw A’p’
btfsc STATUS,Z
goto PWMRoot

PAGESEL MotorControlRoot
movf tmpChar,W
xorlw A’m’
btfsc STATUS,Z
goto MotorControlRoot

PAGESEL TimerRoot
movf tmpChar,W
xorlw A’t’
btfsc STATUS,Z
goto TimerRoot

;————————————————————
; Error

PAGESEL USART_putc
movlw ‘*’
call USART_putc

goto MainLoop

END
[/cc]

Okay, so that’s not quite as good as it could be, but the libraries were better. What about the Java code?

[cc escaped=”true” lang=”java” lines=”20″]
/////////////////////////////////////////////////////////////////////
//
// TACAN – TeleAutonomous Control And Navigation
//
// Mark Dennehy, 93369425
// S.S. C.Eng.
// Final Year Project 1996/7
//
//////////////////////////////////////////////////////////////////////
//
// PolarHistogram.java
// implements the polar histogram
// —————–
// $Date: 1997/04/03 21:49:54 $
// $Revision: 2.0 $
// $State: Stable $
//
//////////////////////////////////////////////////////////////////////

package Tacan.RLC;

import Tacan.util.LogFile;
import Tacan.util.MathUtil;
import java.util.*;
import java.io.*;
import java.awt.event.*;

/**
* The Polar Histogram of the Certainty value grid. This is effectively a
* polar graph of obstacle density.
*
*@author
*@version
*@see Tacan.RLC.CVGrid
*@see java.util.Observer
*@see java.awt.event.ActionListener
**/
class PolarHistogram extends Observable implements Observer, ActionListener
{
private final static double A = 22;
private final static double B = 1;
private final static int maxFreeSectorSize = 18;

private int tmpPH[];
private RLC rlc_;
private int no_of_segments_;
private LogFile log_;
private LogFile data_;
private Vector segment_terms_[];

/**
* The current Polar Histogram values
**/
protected int PH_[];

/**
* The Local Autonomy Control graph values
**/
protected double LAC_[];

/**
* The current maximum value in the Polar Histogram (for normalisation)
**/
protected int PH_max = 0;

/**
* The current maximum value in the Local Autonomy Control graph (for normalisation)
*@see
**/
protected double LAC_max = 0.0;

/**
* The threshold value for deciding what path is free and what is not
**/
protected int PH_threshold = 100;

/**
* The maximum network lag permitted before full local autonomy is granted (in seconds)
**/
public final static double maxNetLag = 10;

/**
* Constructor
*
*@param rlc The Robot Local controller instance associated with this graph
**/
protected PolarHistogram(RLC rlc)
{
rlc_ = rlc;
PH_max = 0;
segment_terms_ = new Vector[rlc.robot_config.PH_RES];
no_of_segments_ = rlc.robot_config.PH_RES;
PH_ = new int[no_of_segments_];
tmpPH = new int[no_of_segments_];
LAC_ = new double[no_of_segments_];
log_ = new LogFile(“Polar.log”);
data_ = new LogFile(“Polar.dat”);

for (int i = 0; i PH_max) ? sum : PH_max;
}
}

public synchronized void LACupdate(int t,int desired_sector)
{
double b = (t LAC_max) ? LAC_[i] : LAC_max;
}
}

public int clearPath(int bearing)
{
int res = (int)360.0/no_of_segments_;
int desired_sector = (int)(bearing/res);
int b;
int l_centre =1;
int l_width = 1;
int r_centre =1;
int r_width = 1;

LACupdate(3,desired_sector);

if ((PH_[desired_sector] * LAC_[desired_sector]) PH_threshold)
{
//Set centre of sector
l_centre = a+(int)(l_width/2);
break;
}
break;
}

//Scan Right
for(int a = no_of_segments_-1; a > (no_of_segments_/2) ; a–)
//Start of first free sector
if ((PH_[(a+desired_sector)%no_of_segments_] * LAC_[(a+desired_sector)%no_of_segments_]) PH_threshold)
{
//Set centre of sector
r_centre = a-(int)(r_width/2);
break;
}
break;
}

//Choose left or right, weighing size of sector with deflection
b = ((l_width/l_centre) > (r_width/(no_of_segments_ – r_centre))) ? l_centre : r_centre ;
b += desired_sector;
b %= no_of_segments_;
b *= res;

return b;
}
}

public String toString()
{
String s = “[ “;
for (int i = 0;iOh dear. That’s very much a curate’s egg. Some parts are okay, but some… well, I wouldn’t give them an okay in a code review session today. Still, I suppose that’s a good thing – remind yourself that you were just not as hot as a teenage coder as you thought you were, and that you’ve gotten better in the interim.

What about the C++ though?

[cc escaped=”true” lang=”cpp” lines=”20″]
/****************************************************************
* *
* TCD Robocup Project 1996/7 *
* Intelligent Agents Group *
* (Artifical Intelligence Group) *
* *
****************************************************************
RCS Block :

$Author: mdennehy $
$Date: 1996/09/07 15:57:18 $
$RCSfile: main.cc,v $
$Revision: 1.6 $
****************************************************************
$Log: main.cc,v $
// Revision 1.6 1996/09/07 15:57:18 mdennehy
// Final version by mdennehy.
// Now knows how to locate itself on the field and
// can determine the error in this measurement.
//
// Revision 1.5 1996/09/04 14:03:22 mdennehy
// Visual and Audio parsing working.
// Minor bug : if more than 5/6 agents connected, Bus errors/segmentation faufollow
//
// Revision 1.4 1996/08/30 17:31:38 mdennehy
// First Working Version
//
// Revision 1.3 1996/08/26 15:10:49 mdennehy
// *** empty log message ***
//
// Revision 1.2 1996/08/24 16:25:08 mdennehy
// Added Threads
//
// Revision 1.1 1996/08/22 14:34:30 mdennehy
// Initial revision
//
****************************************************************/

/***** RCS Program Information *****/
char RCS_revision[]=”$Revision: 1.6 $”;
char RCS_state[]=”$State: Exp $”;

/***** Header Files *****/

#include #include
#include #include
#include
#include
#include
#include

#include
#include
#include
#include
#include
#include
#include
#include

#include “udpif.h”
#include “posdata.h”
#include “communications.h”
#include “dmalloc.h”

/***** Main Comms Socket *****/
UDPSocket comm;

#ifdef DEBUG
/***** Debugging Log file streams *****/
extern ofstream commlog;
extern ofstream datalog;
extern ofstream intsyslog;
#endif

/***** Function prototypes *****/
void usage();
void SignalHandler(int signal, siginfo_t *SignalInfo, void *ContextInfo);

/***** MAIN() *****/
int main(int argc, char **argv)
{
#ifdef DEBUG
print_statistics();
init_debugging_logfiles();
system_check();
intsyslog > s;

// parse returned initialisation message from server
err = sscanf(s,”(init %[lr] %i %[^)]”,side.cstring(),&unum,playmode.cstring());

#ifdef DEBUG
if (err == 3)
{
intsyslog si_code si_uid si_pid si_code == SI_NOINFO)
{
intsyslog si_code)
{
case BUS_ADRALN:
{
intsyslog si_errno si_addr;
intsyslog si_errno si_addr;
intsyslog si_errno si_addr;
intsyslog si_code == SI_NOINFO)
{
intsyslog si_code)
{
case SEGV_MAPERR:
{
intsyslog si_errno si_addr;
intsyslog si_errno si_addr;
intsyslog Erm. Yikes. I did get better though 😀

What about you? Do you check back over long periods to see how your standard of code is actually doing, rather than through the hazy pink-spectacled fog of old memories?

5 comments

  1. Occasionally, but in some cases it’s been so long that either the language has change a *lot* (PHP) or it’s been so long since I’ve coded in the language that I can’t properly understand it any more (C++).

    I still have most of my code from college though, I’m a terrible packrat when it comes to code.

  2. I doubt that my code is really getting better; I don’t think I care much. If my code is already good enough, I’d value productivity improvements and increases in the range of problems that I can solve, over making the code itself a little bit better. When I was a less experienced programmer (but not a beginner) I sweated a lot about code quality, sometimes implementing stuff multiple times to get the interfaces and style just right. Now that I have more experience I can usually come up with something reasonable on the first try, and move on to the next task, rather than spend a lot of time trying to improve the reasonable into the perfect.

    I do remember some code I wrote in college, that took me a day or so to write, for a particular task; recently, I faced the same task again, but I didn’t have the old code, so I wrote a new solution and it took less than an hour. In that way I’m clearly getting better, whether or not the new code is itself better than the old code.

  3. I keep all my code from my teenage years for that reason. Just to laugh at my past self.

    That PIC assembly brought some memories back. I used to do some PIC programming as well… in assembly (http://viennot.biz/remote_car.pdf / page 33). It’s only later that I discovered what was a C compiler.

    Turns out, what most of us need to get rid of comments is to create our own DSLs, but that’s hard. What’s even worse is that most of the time, trying to get there is pure intellectual masturbation.

  4. I noticed your mention of base-lining in that discussion, was wondering if you actually did that yourself 🙂

    I don’t code all that much myself these days so I reckon I’m probably on a down curve (or at best seeing a significant reduction in the rate of improvement). I need to get stuck in a bit to keep my head in.

    Looking back at my code from college, what’s most interesting to me is the world of difference between what I wrote in 3rd year, and what I wrote in 4th year after working for a software house for 12 months. The code isn’t necessarily more advanced – I worked in 360/390 ASM for the 12 months, which wasn’t on our course – but the primary difference is the defensive coding and documentation. Designed for robustness and maintenance.

    I’d be interested to know what the nature of the code improvements are that most people get from their first couple of years in industry – is it totally dependent on the company, or the language/environment? Or something else?

  5. Nice post. Thanks.

    I do keep my old code, but I rarely look at it. This is a good idea I’ll start doing it.

    I do try and benchmark my self. I will do various coding exercises and look at the interview process for big companies. I also look at websites like the one below and see how I compare. Through the years I have “leveled up”, but as expected my advancement is slowing down as I get more specialized.

    http://www.indiangeek.net/programmer-competency-matrix/

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.