[Translation] Obocchama Kun

[MooZ]

I'm currently fighting against the printing routine. It's a mess...
I think I'll end up rewriting it completely. Maybe I'll be able to use BPE instead of DTE. The gain of DTE is rather small. The complete french script is 1732 bytes long. The DTE script is 901 bytes long, but the associated dictionary is 466 bytes long. This saves 365 bytes...
Anyway! Here's the mandatory preview screenshot:

[MooZ]

One year later...

The introduction and monster names are done. I'm currently fighting with the end credits. As you may have guessed, the text reading routine is different from the introduction/monster one.
It's simpler. It's just:

Code: Select all

b66e:    ; read a char and display it, stop if it's #$ff
        lda [$20]    
        cmp #$ff
        beq b67c    
            jsr b6a1    ; this is where the BAT pointer is incremented and the char added to the "display list"
            jsr b687
            bra b66e
b67c:   
        jsr b687    
        lda [$20]    
        sta <$22      
        jsr b687    
        rts          
        
b687:    ; increment text pointer
        clc          
        lda <$20      
        adc #$01     
        sta <$20      
        cla          
        adc <$21      
        sta <$21      
        rts

Unfortunately, the text is compressed. I can't just do a lda [$20]. I have to call the decompression routine. Hopefully space can be saved by rewriting the previous code a little bit.

Code: Select all

b66e:
        lda [$20]    
        cmp #$ff
        beq b67c    
            jsr b6a1
            jsr b687
            bra b66e
b67c:   
        jsr b687    
        lda [$20]    
        sta <$22      
        ; As the increment routine follows, we can brutally the jsr/rts and save 4 bytes.
b687:
        inc <$20      
        bne .inc_end          
            inc <$21      
inc_end:     
        rts

10 bytes saved!

The next thing to do is to remove the simple loads with the decrompression routine. The thing is that the decompression routine automatically increment the text pointer. We don't need to call b687 in the read loop.

Code: Select all

b66e:
        jsr bb1a    ; decompress text
        cmp #$ff
        beq b67c    
            jsr b6a1
            bra b66e
b67c:
        lda [$20]      
        ; The rest is left unchanged

We removed 2 jsr but changed the lda <zp with a jsr. This makes a 5 bytes save.
We now have 15 bytes of free space. This will come handy because as you may have noticed, at the end of the text reading we read an extra byte and store it in <$22.
This byte seems to be the y BAT position of the text. This means that I just can't brutally encode the text as I did for the introduction. I need to modify the script and its encoder to add this extra byte. Otherwise the whole display is foobared (and it currently is).
And by looking at the ROM, it's not just 1 extra byte but 2 bytes. One at the end of the string and another one at the beginning of the next one.
I'll have to check the b66e caller...
Yeepee!