Better code check before compile?

[pedja089]

Then 80% of my programs wouldn't compile. I often use multiple entry point in same sub. And have only one return at end.

[Ioannis]

Or the reverse. Have one entry and multiple exits:

gosub test

....

test:
if x=1 then
   a=1
   return
endif

if x=2 then
   a=2
   return
endif

return

Ioannis

[CuriousOne]

So you're doing it wrong - same for overlapping for-next cycles

[Ioannis]

If you mean my example, OK, it is a not efficient one but I was trying to show that you can have many Returns from the same sub.

Sure it can be made more compact or efficient, but still. Compiler has difficult time to check this for errors.

Ioannis

[CuriousOne]

Ok, here is an example (not exact syntax, just to get idea)

MAINCODE: 'do something
high PORTC.0
PAUSE 500
low PORTC.0
Pause 500
GOSUB CHECKER
GOTO MAINCODE
CHECKER:
TOGGLE PORTD.0
RETURN

In above example, if we remove RETURN, it will compile fine in PBP, but will produce unusable code. But even on ZX Spectrum, it would give us "Gosub without return" error, and save our time.

[Ioannis]

Agreed. This is easy to check, but above we stated other cases that are difficult to check.

So a sub-without-return check in whole is not easy to implement. I suppose... Maybe Charles knows better to justify it.

Ioannis

[mpgmike]

There have been situations where if a test meets certain criteria you want to go through 5 sequences of code. However, if one of the criteria is missing, you want to skip the first sequence and do the other 4. On down the line, you may only need to do the last 2. The subroutine may look like;

Main:
SELECT CASE Var
 CASE < 50 : GOSUB Sub1
 CASE < 100 : GOSUB Sub2
 CASE < 150 : GOSUB Sub3
 CASE < 200 : GOSUB Sub4
 CASE ELSE : GOSUB Sub5
END SELECT
GOTO Main

Sub5:
 TOGGLE LED1
Sub4:
 TOGGLE LED2
Sub3:
 TOGGLE LED3
Sub2:
 HIGH LATA.4
Sub1:
 LATB = 0
RETURN

Here is a legitimate case of needing only 1 RETURN for multiple subroutines.