Feature request - Exit Case/Break

[Unseen Machine]

Wow, if only all my threads got such a response!

Anyway, I only asked cause I came across the need for it, admittedly it's only the once in 10 years of coding in QB64 i've had this need though so i suppose i could just use a work around.

Here's a basic view of how i came across the need for _BREAK. I had a select case with a triple For loop in it and rather than the work around of checking a variable and then exiting each for loop i needed a _BREAK.

Code: QB64: [Select]

1. SELECT CASE i&
2.  
3.   CASE 1
4.     FOR k% = 0 TO 10
5.  
6.       FOR j% = 0 TO 10
7.  
8.         FOR i% = 0 TO 10
9.  
10.           if K% = Col% and j% = row% and i% = i& then _BREAK
11.  
12.         NEXT
13.  
14.       NEXT
15.  
16.     NEXT
17.  
18. END SELECT

Obviously i could use Line labels and a GOTO (and is what i ended up doing) but thats not (as far as i am concerned) good coding practice, that maybe just a hangup i have being a OOP guy or something, i dunno.

I dont mind if the answers a NO as it's far from essential.

Unseen

[Dimster]

This topic touches on a How To question for me.  If you have a Select Case within a Do Loop and you want to code one of those Cases to get you out of the Selection process and back into the Do Loop does _Continue do that? or does _Continue just skip the Case it's found in a moves onto the next Case?

[FellippeHeitor]

_CONTINUE goes to the beginning the loop, regardless of whether you're inside a select case block or not.

That, of course, if the condition for looping is still true. Otherwise, the loop block is exited.

[FellippeHeitor]

Steve's remarks considered, I believe we're aiming for EXIT SELECT and EXIT CASE, for when EVERYCASE is used. Does anybody have anything else to add?

[Cobalt]

I'm afraid I must disagree.

My argument is we already have a way, however despise-able it is, to jump around in a program.

This would be nothing more than an alias of GOTO.

btw what is wrong with this exactly?

Code: QB64: [Select]

1.     SELECT CASE i&
2.      
3.       CASE 1
4.         FOR k% = 0 TO 10
5.      
6.           FOR j% = 0 TO 10
7.      
8.             FOR i% = 0 TO 10
9.      
10.               IF K% = Col% AND j% = row% AND i% = i& THEN MatchFlag%% = -1: i% = 11: j% = 11: k% = 11
11.      
12.             NEXT
13.      
14.           NEXT
15.      
16.         NEXT
17.      
18.     END SELECT

[FellippeHeitor]

The good thing is that there's no new keyword added to the language. Just a new way to use them, which won't hurt anyone at all.

My point when I ask if anyone has anything to add at this point is anything technical regarding the implementation as it's been showcased above.

[bplus]

Yes sounds very good to me, just extending where you can apply EXIT, no new keywords.

[SMcNeill]

The good thing is that there's no new keyword added to the language. Just a new way to use them, which won't hurt anyone at all.

My point when I ask if anyone has anything to add at this point is anything technical regarding the implementation as it's been showcased above.

Only technical point is to be careful of flow into a CASE ELSE block of a SELECT EVERYCASE.  Make certain it’s working as intended.

As long as we’re implementing this, can we also get an EXIT <counter> command?

FOR I
   FOR J
       FOR K
           EXIT 2
       NEXT
   NEXT
NEXT

The above would exit 2 control blocks (K and J blocks) and flow down into the I block, after that 2nd NEXT.

Or maybe an EXIT TO label statement?  That could be useful too.


FOR I
   FOR J
       FOR K
           EXIT TO foo
       NEXT
   NEXT
foo:
NEXT

[FellippeHeitor]

Anyone willing to test the new feature, please head to the development build page and get the latest version:

https://www.qb64.org/portal/development-build/

EXIT SELECT
Can be used to exit a SELECT CASE block.

EXIT CASE
Has the same purpose of EXIT SELECT, unless when used in a SELECT EVERYCASE block. In such case, it'll skip to the next CASE evaluation.

Please try to break it. I'll be watching this thread.

PS: Anyone new to development builds, here's the changelog since v1.4 was released: https://www.qb64.org/portal/development-build-changelog/

[FellippeHeitor]

As long as we’re implementing this, can we also get an EXIT <counter> command?

I wouldn't oppose you implement it.

Or maybe an EXIT TO label statement?  That could be useful too.


FOR I
   FOR J
       FOR K
           EXIT TO foo
       NEXT
   NEXT
foo:
NEXT

But this is definitely a GOTO.

[Unseen Machine]

https://www.qb64.org/portal/development-build-changelog/

I couldnt see the _IFNOTINCLUDED function i previously requested here https://www.qb64.org/forum/index.php?topic=2733.msg119540#msg119540

Did this not get implemented? Is it in the works?

Thanks for putting this together though, now i've just gotta figure out what it was i was coding when i needed it!

Unseen

[FellippeHeitor]

The trick to check if a precompiler setting is undefined was already there. You just didn't know about it. Have you started using it for your libraries?

[Richard]

Trying to "break"  exit case

Code: QB64: [Select]

1.  
2. a$ = CHR$(0)
3. SELECT EVERYCASE a$
4.     CASE CHR$(0):
5.         'EXIT SELECT
6.         : EXIT CASE
7.     CASE CHR$(1)
8.         BEEP: BEEP
9.     CASE ELSE
10.         BEEP: BEEP: BEEP: BEEP: BEEP
11. END SELECT
12. BEEP
13.  
14. END
15.  

using Dev version f3b92c3  Windows 10 x64

I thought 3 x BEEP was to be the result.


HOWEVER

Code: QB64: [Select]

1. a$ = CHR$(0)
2. SELECT EVERYCASE a$
3.     CASE CHR$(0):
4.         'EXIT SELECT
5.         : EXIT CASE
6.     CASE CHR$(1)
7.         BEEP: BEEP
8.     CASE CHR$(0)
9.         BEEP:
10.     CASE ELSE
11.         BEEP: BEEP: BEEP: BEEP: BEEP
12. END SELECT
13. BEEP
14.  
15. END
16.  

gives 2 x BEEP which is logically correct.

[SMcNeill]

Trying to "break"  exit case

Code: QB64: [Select]

1.  
2. a$ = CHR$(0)
3. SELECT EVERYCASE a$
4.     CASE CHR$(0):
5.         'EXIT SELECT
6.         : EXIT CASE
7.     CASE CHR$(1)
8.         BEEP: BEEP
9.     CASE ELSE
10.         BEEP: BEEP: BEEP: BEEP: BEEP
11. END SELECT
12. BEEP
13.  
14. END
15.  

using Dev version f3b92c3  Windows 10 x64

I thought 3 x BEEPS was to be the result.

Shouldn’t it be a single beep?  The first case is the only one that matches, and it doesn’t beep.  The only beep that I’d think should be triggered, is the one after the END SELECT.

[FellippeHeitor]

Code: QB64: [Select]

1.  
2. a$ = CHR$(0)
3. SELECT EVERYCASE a$
4.     CASE CHR$(0):
5.         'EXIT SELECT
6.         : EXIT CASE
7.     CASE CHR$(1)
8.         BEEP: BEEP
9.     CASE ELSE
10.         BEEP: BEEP: BEEP: BEEP: BEEP
11. END SELECT
12. BEEP
13.  
14. END
15.  

The above beeps once.

Code: QB64: [Select]

1. a$ = CHR$(0)
2. SELECT EVERYCASE a$
3.     CASE CHR$(0):
4.         'EXIT SELECT
5.         : EXIT CASE
6.     CASE CHR$(1)
7.         BEEP: BEEP
8.     CASE CHR$(0)
9.         BEEP:
10.     CASE ELSE
11.         BEEP: BEEP: BEEP: BEEP: BEEP
12. END SELECT
13. BEEP
14.  
15. END
16.  

The above beeps twice.

All as expected so far.