Suggestions on formatting of code

[PeterJ]

Hi,

Can I ask about formatting code?

I appreciate the 'proper' way of doing it is like the example below (This example of for InkSpector):

Code: Select all

	TARGET ZXSPECTRUM48
	SAVETAP "sc.tap"
	ORG	$6000

main:	ld a,2
	call 5633
loop:	ld de,string
	ld bc,eostr-string
	call 8252
	jp loop

string defb 22,10,10,17,5,72,69,76,76,79,32,87,79,82,76,68,33
eostr  equ $

	END main

However, Pasmo is far more relaxed about formatting, and this runs just fine:

Code: Select all

org $6000
main
ld a,2
call 5633
loop
ld de,string
ld bc,eostr-string
call 8252
jp loop

string defb 22,10,10,17,5,72,69,76,76,79,32,87,79,82,76,68,33
eostr  equ $

END $6000

I understand that the first line is much better in terms of readability, but having to tab after every line is a bit of a pain.

Any thoughts from others on this?

[chilledgamer]

Of course I am new and cannot comment on proper formatting, but just thought I would let you know something I found out recently.

I am now using VSCode. and I can highlight as many lines as I want with the mouse and press tab, and it just works . I'm sure you already new this , but just in case :d

[Einar Saukas]

Properly organizing and documenting source code is done for your own convenience, not for the assembler/compiler.

You don't HAVE to do it. But good luck trying to do something useful without doing it

[PeterJ]

I'm all for documenting [mention]Einar Saukas[/mention]!

Thanks [mention]chilledgamer[/mention], I use that with Python on Visual but just tend to use notepad for my z80.

It's just interesting how some assemblers actually force good formatting on you.

[Dr beep]

I would not only go for the outlined version for readability, but also add all kinds of comment so after years you still can figure out what it exactly does.

To give you an example, I have a project, started 2011, bug found after 2 years and took me untill 2016 to find what went wrong.
Code itself is perfect but the stack of records grew over the program. i had to replace piece of code. Why code is placed at #6000 and not At #8000 is now explained in commentlines.

[Seven.FFF]

Sometimes it is for the assembler. Some assemblers have rules that anything starting in column 0 is a label, and anything in columns 1+ is an opcode. Some assemblers allow reserved words as labels without trailing colons provided they are in column 0.

Basically the simpler the parser, the more likely you are to break its assumptions and produce uncompilable or wrong code if you don't follow its formatting rules. There are some real doozies out there with simplistic parsers.

[PeterJ]

Just to add, I do normally include comments. I didn't in my sample as I didn't want to loose focus from the formatting angle. Cheers.

Thanks [mention]Seven.FFF[/mention]. Pasmo must be pretty clever to workout my dummy coding!

[Morkin]

One suggestion that I found useful when I was using ZX Spin is to always put a colon (:) at the end of labels, so I could use CTRL+F to find them (including the colon).

Without the colon, once the code got quite big I was encountering loads of matches, e.g. in LD statements, before I got to the label itself when I did CTRL+F.

Other assemblers may have ways to find labels in another way, but after a while it's easy to end up with hundreds of labels, so even an alphabetical list takes a while to scroll through especially if a lot of labels are similar.

[PeterJ]

Thanks [mention]Morkin[/mention],

I know a lot of people quite rightly love Spin, but because it's so long in the tooth now I have problems with it in W10 (I know it's fine for some users).

At the sort of length code I do, notepad and Pasmo is fine for me, but that functionality with labels is useful. I will start using colons though so I can find labels in notepad.

Maybe one day we will get Spin V2 [mention]ZXDunny[/mention]? Here's hoping.

[ketmar]

mine is usually like this. labels are always on a separate line, and instruction arguments are always at column 8. in-line comments are either at column 16 (if it fits), but almost always aligned.

some code snippets

Code: Select all

;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
;; build fully unrolled screen blitter
;; generates 4329 bytes of code
;; blitter timing is 35897, so we can avoid DI/EI
;; in:
;;   nothing
;; out:
;;  all registers are dead
;;  all alternate registers are dead
;;
VScrBuildBlit:
  ld    de,CodeEnd

  exx

  ; start of vscr
  ld    hl,V_ADDR+1

  ; start of scr$+width/2
  ld    bc,scraddr8x8(4,1)+VSCR_WIDTH/2-1

  ; start of attr+width/2
  ld    de,scrattr8x8(4,1)+VSCR_WIDTH/2-1

  exx

  ; put prologue
  ld    hl,blitjit_prologue
  ld    bc,blitjit_prologue_len
  ldir

  ld    a,VSCR_HEIGHT

.jitnext9:
  ex    af,af'

  ; put scr$ copy parts (8 lines)
  ld    b,8

.jitnextvline:
  push  bc

Code: Select all

;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
;; draw sprite with 16x8 tiles, with full-tile clipping
;; in:
;;  B: y (chars)
;;  C: x (chars)
;;  HL: sprite data
;; out:
;;  all main registers are dead
;;  AF': dead
;; visible vscr starts at (4,1)
;; size: (24,12)
;;
DrawSpriteSimple16xn:
  ; horizontal clip
  ld    a,c
  cp    VSCR_WIDTH
  ret   nc
  ; vertical clip
  ld    a,b
  cp    VSCR_HEIGHT
  jp    nc,.checkneg
  ld    a,(hl)  ; block counter
  add   a,b     ; end y
  sub   a,VSCR_HEIGHT+1
  ld    a,(hl)  ; most of the time it will be ok
  ; if the result is positive (or zero), we're ok
  jp    m,.vertok
  ; fix counter
  ld    a,VSCR_HEIGHT
  sub   b
.vertok:
  ex    af,af'

[PeterJ]

but almost always aligned.

Thanks for the sake of my OCD, it must be aligned always!

[ketmar]

but almost always aligned.

Thanks for the sake of my OCD, it must be aligned always!

and it is also easier to quickly scan the code too. syntax highlighting helps too, but with aligned parts it is even easier. this habit (aligning) actually came from GENS assembler (it did it automatically for you).

[catmeows]

Basically what everybody says:
1) comment, comment, comment ...
2) align by columns
3) meaningfull labels

Code: Select all

dutyModification
	ld a, (ix + DUTY)		
	ld c, (ix + DUTYDELTA)
	add a, c			;add duty change to duty
	bit 7, c			;is delta negative ?
        jr nz, negDutyDelta
	ld b, (ix + DUTYMAX)		;compare with max duty
	cp b
	jr c, noDutyChange		;will not change duty delta sign
dutyChange
	ld a, c				;negate duty delta			
	neg 
	ld c, a
	ld a, b				;and set min/max limit as duty
	ld (ix + DUTYDELTA), c		;store new delta
noDutyChange
	ld (ix + DUTY), a		;store new duty
	ret
negDutyDelta
	ld b, (ix + DUTYMIN)		;compare with min duty
	cp b				
	jr nc, noDutyChange		;will not change duty delta sign
	jr dutyChange

noteModification
	ld a, (ix + NOTELEN)		;decrease note length
	dec a
	ld (ix + NOTELEN), a
noteNoLastTick
	ld a, (ix + ARPFLAG)		;is appergio enabled ?
	or a
	jr z, noteNoArp			;no
	ld a, (tickCounter)		;if appergio enabled
	and 1				;every second tick play tone one octave higher
	jr z, noteNoArp
	ld a, 12			;octave plus
noteNoArp
	add a, (ix + BASENOTE)		;add nothing or octave to base note
	call getNoteFreq		;get note frequency
	ld a, (ix + VIBRATOFLAG)
	or a
	jr z, noteNoVibrato		;no vibrato		
	ld de, 47			;1 + 11*2 = 23
	add hl, de			;move note 
	ld a, (hl)
	inc hl
	ld h, (hl)			;get frequency for note+1
	ld l, a
	or a
	sbc hl, bc			;difference between notes to HL	
	ld a, (ix + VIBRATODEPTH)		
vibratoDepth	
	srl h
	rr l
	dec a
	jr nz, vibratoDepth		; delta of frequency / ( 2^vibratoDepth)
	ld a, (vibratoCounter)		; now multiply vibrato depth by vibracounter
	or a
	jr z, noteNoVibrato		;level 0
	ld d, h				
	ld e, l				
	bit 1, a				
	jr nc, vibratoBit1		
	add hl, hl			
vibratoBit1
	rra
	jr c, vibratoBit0
	add hl, de
vibratoBit0
	add hl, bc			; add vibrato to base note frequence
	ld c, l
	ld b, h
noteNoVibrato	
	ld a, (ix + PORTAMENTOFLAG)
	or a
	jr z, noteNoPortamento		;no portamento 
	ld e, (ix + PORTAMENTOLO)
	ld d, (ix + PORTAMENTOHI)
	ld l, e
	ld h, d
	add hl, bc			;add portamento to frequency
	ld c, l				;store final frequency
	ld b, h
	ld l, (ix + PORTAMENTODELTALO)	
	ld h, (ix + PORTAMENTODELTAHI)
	add hl, de			;update portamento
	bit 7, (ix + PORTAMENTODELTAHI)	;is delta negative ?
	jr z, notePortamentoChange
	ccf
notePortamentoChange
	jr nc, notePortamentoContinue
	xor a				;switch portamento off
	ld (ix + PORTAMENTOFLAG), a
notePortamentoContinue
	ld (ix + PORTAMENTODELTALO), l	;store portamento delta
	ld (ix + PORTAMENTODELTAHI), h
noteNoPortamento
	;BC is final frequency
	ld a, (ix + RELEASEFLAG)
	or a
	ret z				;if not release then exit
	ld a, (ix + NOTELEN)
	or a
	ret nz				;is last tick ?
	ld bc, 0			;erase frequency for the very last tick
	ret

[Ralf]

I do both ways.
When coding for Spectrum I never use automatic formatting, I always do it my own, (sometimes inconsistent ) way.

When I have short labels I May do

Code: Select all

DS1 LD A,(HL)
    INC HL
DS2 LD B,(HL)

But what if I have a label like CopyEnemyGraphicDataToBuffer ?
And yes, I use this kind of labels too, we may have this comfort, it's 2020.

Imagine typing these all spaces/tabs with each instruction:

Code: Select all

CopyEnemyGraphicDataToBuffer LD A,(HL)
                             INC HL
                             LD B,(HL)

So in case of long labels I prefer

Code: Select all

CopyEnemyGraphicDataToBuffer 
 LD A,(HL)
 INC HL
CopyEnemyGraphicDataToBuffer 2
 LD B,(HL)

[Ralf]

I would just strongly advise keeping komments aligned:

LAZY AND BAD:

Code: Select all

LD C,D ;comment1
LD A,(PlayerStatus) ;comment2

GOOD:

Code: Select all

LD C,D              ;comment1
LD A,(PlayerStatus) ;comment2

[PeterJ]

I know it doesn't make a jot of difference to the assembler, but it's interesting how some of you are using uppercase, and others lowercase.... Just be consistent I suppose is the answer.

[Seven.FFF]

Again it may make a difference, as some assemblers will treat a reserved word as an opcode if it’s entirely upper or lower case, but allow the same reserved word as a label if it is in mixed case.

Also some editors and IDEs detect the intent of the current line, and will auto-indent the next line by the same amount when you hit enter, so having a large indent optimised for long labels doesn’t cause more effort.

[Ralf]

I know it doesn't make a jot of difference to the assembler, but it's interesting how some of you are using uppercase, and others lowercase.... Just be consistent I suppose is the answer.

It's called S T Y L E

[Stefan]

IN THE OLD DAYS UPPERCASE WAS USEFUL SINCE LOWERCASE LETTERS COULD BE HARDER TO READ, THESE DAYS, UPPERCASE IS JUST CONSIDERED SHOUTING!!!!

(yes I grew up from uppercase to lowercase).

[kolbeck]

https://youtu.be/SsoOG6ZeyUI

[PeterJ]

Excellent [mention]kolbeck[/mention]

I haven't laughed so much in months.

[andydansby]

My preference has been for camelCase, just easier on my eyes.

[kolbeck]

Excellent @kolbeck

I haven't laughed so much in months.

One of my favorite episodes

[ketmar]

as for me, i just can't remember how to type those big letters. have to copy-paste them from a premade text file.

[Ast A. Moore]

Okay, the video was brilliant!

Now, to address some of Peter’s concerns. Many text editors that are tailored to programmers will do a few things for you to make your life easier. For one, they provide syntax highlighting (not the editors themselves, but their extensions). Take a look at the one I wrote for TextMate:



Second, TextMate (and many others) will automatically tabulate your code. Once you’ve started a line with a tab, when you press Return, the next line will be automatically tabulated to line up with the line before. I don’t know what text editor you use, but I know that the reason I switched to TextMate from TextEdit (the macOS version of Notepad) and wrote the Z80 syntax highlighter was precisely because of these two problems: code readability and tabulation.