Text Only | Text with Attachments

QB64.org Forum

Active Forums => QB64 Discussion => Topic started by: Cobalt on June 26, 2021, 11:46:01 am

Title: Confirmation of proper behavior in SELECT\CASE
Post by: Cobalt on June 26, 2021, 11:46:01 am
Ok I think I know what is going on here but I just wanted more input on this,

Is this how SELECT\CASE in QB64 is supposed to handle this?
Code: QB64: [Select]
  1. DIM TEST AS _BYTE
  2. TEST = &HFF
  3. 'TEST = &H80
  4. 'TEST = &H40
  5. 'TEST = &H08
  6. 'TEST = &H01
  7.  
  9.  CASE &HFF 'flaged conversations
  10.   PRINT "FF is GTG!"
  11.  CASE &H80
  12.   PRINT "80 is GTG!"
  13.  CASE &H40 'shops
  14.   PRINT "40 is GTG!"
  15.  CASE &H08 'church
  16.   PRINT "08 is GTG!"
  17.  CASE &H01 'standard dialog
  18.   PRINT "01 is GTG!"
  20.  
  21. PRINT "&H"; HEX$(TEST)
  22. PRINT TEST
  23. C~%% = TEST
  24. PRINT C~%%
  25. T%% = &HFF
  26. PRINT T%%
Its not catching the first 2 CASE scenarios here. least not on my side, V1.5 stable

I think it may treating the &H as an unsigned value?(&H80 and up sets bit 7 which makes it a negative in a signed scenario)
should it be treating the values differently? if you assign TEST to an unsigned variable it doesn't change it, the BITs are still set properly so why isn't SELECT\CASE matching it? if you put &HFF into a signed variable it becomes -1.

From my perspective &HFF is both -1 and 255 based on the SELECT variable used; SIGNED or UNSIGNED.

Its an easy fix of adding _UNSIGNED to my UDT in the actual code, but I just wanted to make sure the SELECT\CASE was behaving properly, which I think it might be but I'd rather have input from others on that one to make sure.
Title: Re: Confirmation of proper behavior in SELECT\CASE
Post by: SpriggsySpriggs on June 26, 2021, 12:44:35 pm
I'd just make TEST an unsigned byte because when you are doing the case, &HFF is going to be pointing to 255 while your variable will be overflowed and not match the value. That's my guess
Title: Re: Confirmation of proper behavior in SELECT\CASE
Post by: SMcNeill on June 26, 2021, 12:53:38 pm
Code: QB64: [Select]
  1. Dim TEST As _Byte
  2. TEST = &HFF
  3. 'TEST = &H80
  4. 'TEST = &H40
  5. 'TEST = &H08
  6. 'TEST = &H01
  7.  
  9.     Case &HFF%% 'flaged conversations
  10.         Print "FF is GTG!"
  11.     Case &H80%%
  12.         Print "80 is GTG!"
  13.     Case &H40%% 'shops
  14.         Print "40 is GTG!"
  15.     Case &H08%% 'church
  16.         Print "08 is GTG!"
  17.     Case &H01%% 'standard dialog
  18.         Print "01 is GTG!"

Assign the variable type to your &H-math manually, like I've done above, and QB64 won't have to guess at what that value is supposed to be.
Title: Re: Confirmation of proper behavior in SELECT\CASE
Post by: Cobalt on June 26, 2021, 06:10:23 pm
Quote from: SpriggsySpriggs on June 26, 2021, 12:44:35 pm
I'd just make TEST an unsigned byte because when you are doing the case, &HFF is going to be pointing to 255 while your variable will be overflowed and not match the value. That's my guess

Yeah, I mean that is what I ended up doing.  QB64 defaults to LONG with undefined variables and constants or SINGLEs? so &HFF is probably being treated as a LONG or SINGLE even though the comparison variable is BYTE. I'd almost think it would want to match the CASE type with the SELECT type.
In terms of BITs they are identical,at BYTE width, &HFF = 11111111, -1=11111111, and 255=11111111. So in a bit-wise comparison they should always be =.
Been doing a lot of BIT work in my game lately, so probably why I was taken aback by this.

Quote from: SMcNeill on June 26, 2021, 12:53:38 pm
Assign the variable type to your &H-math manually, like I've done above, and QB64 won't have to guess at what that value is supposed to be.

Oh yeah, I forgot about that. Been a long time since I have used that method.


✅Thanks for the feedback guys.✅
Title: Re: Confirmation of proper behavior in SELECT\CASE
Post by: NOVARSEG on June 27, 2021, 02:31:22 am
I looks like  the way CASE is interpreting &HFF

Code: QB64: [Select]
  1. DIM TEST AS _BYTE
  2. TEST = &HFF
  3. PRINT TEST
  4.  
  5. 'TEST = &H80
  6. 'TEST = &H40
  7. 'TEST = &H08
  8. 'TEST = &H01
  9.  
  11.     CASE -1 'flaged conversations
  12.         PRINT "FF is GTG!"
  13.     CASE &H80%%
  14.         PRINT "80 is GTG!"
  15.     CASE &H40%% 'shops
  16.         PRINT "40 is GTG!"
  17.     CASE &H08%% 'church
  18.         PRINT "08 is GTG!"
  19.     CASE &H01%% 'standard dialog
  20.         PRINT "01 is GTG!"
  22.  
Text Only | Text with Attachments