Hi

In the normal basic of zx spectrum 48k there is a behaviour or trick like this

you can change the address of DEFADD variable

and after this you can copy bytes very fast  Is like an assembler LDIR

I will explain more

you can have 2 variables A$ and B$  for example

every one of this variables can have 1000 bytes

when you do A$ = B$  the 1000 bytes of B$ are copied on A$

Normally DEFADD points to the beginning of the variables definitions

for example  (invented numbers)

DEFADD_POINTER_ADDRESS    ;;'' defadd initial definitions of variables
A$ begins on 30000 and have 1000 bytes of size
B$ begins on 32000 and have 1000 bytes of size
etc ...


The trick is that in BASIC we can change the DEFADD address and put any thing

for example we can do

CHANGED_DEFADD_POINTER_ADDRESS    ;;'' CHANGED defadd initial definitions of variables
A$ begins on 16384 and have 256 bytes of size
B$ begins on 32000 and have 256 bytes of size
etc ...


This way we can do A$ = B$ and copy 256 very fast to the screen !!!

this is better explained here in spanish
https://blog.jafma.net/2020/03/16/effici...m-iv/#sp_5

and here in english
https://blog.jafma.net/2020/03/16/effici...m-iv/#en_5

and here a video in spanish explaining this trick
https://www.youtube.com/watch?v=VKJ2dePykdA

I tried this trick with compiled zx basic but is not working because seems
that string variables are not doing the same behavior that original ROM

Hi

Is there a list somewhere that explains how to optimize?

i  explain

Normally in compilers if you put the code in one way it is more efficient than if you put the code in another way

For example you can do

variable ++    or   variable = variable + 1

or maybe if you use local variables it is more efficient than if you use global variables

perhaps the compiler translates with fewer bytes and more efficiently one way or another

Is there somewhere a recomended list of tricks to write a more efficient syntax that translates better to assembler

and uses less memory and fewer bytes?

Hi  if i compile this, says

warning: [W180] Unreachable code

but i don't know why

the version is  zxbasic-1.15.2

why is there 2 halt on waitretrace macro?

i found in the library this waitretrace macro in the  retrace.bas  file

REM simple WaitRetrace macro

#ifndef waitretrace
#define waitretrace 'REM Retrace \
asm \
halt \
halt \
end asm

#endif


Is not better put only one halt for better timer control of the programmer?

if there was only one halt is more easy to control

with one halt we can do

waitretrace  ->  only 1 halt  1/50 seconds

2 halt      -> 2/50seconds
waitretrace   
waitretrace    

N halt    -> N/50seconds
for i = 1 to N
   waitretrace
next i

Hi 

In assembler there is a register that have the state of the flags zero, carry , etc

It is possible to read the state of the flags ONLY WITH BASIC (NOT ASSEMBLER) or the state of the carry flag, zero flag etc

Hi

Is posible put a ubyte variable just after DIM   WITHOUT ASSEMBLER  maybe with AT @  or something

Example:

Trying to use CHR$(13) in print42() and it is ignored - could this be implemented please?


Thanks

since a long time i was thinking how simple or possible would be using sprite engines/libraries on zxbasic-compiler (just like those used on z88dk and so on)

so i started to try this (very glitchy and inefficient, as the code shows...   )

(wsad or cursors for moving)

I wonder if you could show some examples of how to use #include MACRO introduced in 1.5.1 as I don't see anything in the wiki

thanks

recently i was trying this
https://worldofspectrum.org/forums/discu...n-assembly
is there some more efficient way, that works?
all my examples there were totally made on ZX-Basic Compiler - i'm now stuck in the last one, i only can load one file when it was supposed to load more...