Strange interaction between long variable and HSerPrint of concatenated value

[Jim giordano]

When I do an HSerPrint of a concatated string, and also do an operation with a variable declared as Long, there is some sort of interaction that trashes the values in one of my other arrays.

example:

#option Explicit
#chip 16LF1938, 32

#define USART_BAUD_RATE 115200 '460800 '460800 '921600 '230400 '115200
#define USART_TX_BLOCKING

#define prt HSerPrint

dim seg(9)
dim ii
dim timel,timeh
dim bigtime as long

prt "test"

for ii=1 to 8:seg(ii)=63:next ii  'preset all values to 63

prt chr(27):prt "[5;1H"
for ii=1 to 8
  prt seg(ii):prt " "  ' print out values before the concatenation
next i

prt chr(27):prt "[3;1H"
prt "abcde"+"fghij"  ' this caused the problem

prt chr(27):prt "[6;1H"
for ii=1 to 8
  prt seg(ii):prt " "   ' print out values after the concatenation
next i
do
loop

' the rest is the original program stripped down to the
'   minimum that still affects the problem
timel=TMR1L
timeh=TMR1H-0x80 ' subtract out starting amount
bigtime=timeh*256+timel   ' ** this instruction causes the problem

Note that if I comment out the last line, the problem doesn't occur, even though it is never executed.

The result look like:

°°est

abcdefghij

63 63 63 63 63 63 63 63
103 104 105 106 63 63 63 63

so you can see that the first four values of the seg array were changed to "ghij"

[Anobium]

I have seen this before. Try.

hsersend 27: wait 100 us:prt "[5;1H"

I think it is the terminal processes the <esc>. </esc>

My test on a 16f1938

test

abcdefghij

63 63 63 63 63 63 63 63
103 104 105 106 63 63 63 63

[Anobium]

and, you can use... Timer1 and cut that code.

bigtime=Timer1

[Jim giordano]

that's two interesting things, but it has no effect of what is trashing the seg array.

What interaction would change the first four values in and unrelated array.

It seems the compiler is overlapping a declared array with some of it's intermediate values, not a good thing.

[Anobium]

Is it? I think not.

Where is it in my response? I dont understand the issue - me thinks.

What version of compiler etc. Look at the first line of the ASM

[Jim giordano]

;Program compiled by Great Cow BASIC (0.98.06 2019-06-12 (Windows 32 bit))

I guess I'm missing something here.

This is the seg array before the print -
63 63 63 63 63 63 63 63

and after doing - HSerPrint "abcde"+"fghij"

here is the seg array after that print-
103 104 105 106 63 63 63 63

Why would an HSerPrint change the values in an unrelated array?

Notice that the values changed were changed to 103,104,105,106 which is ascii for "ghij"

So the compiler put together code that used the first part of my array for scratch.

From the asm file-

;prt "abcde"+"fghij"  ' this caused the problem
  movlw low SYSSTRINGPARAM1
  movwf FSR1L
  movlw high SYSSTRINGPARAM1
  movwf FSR1H
  clrf  SysStringLength
  movlw low StringTable5
  movwf SysStringA
  movlw (high StringTable5) & 127
  movwf SysStringA_H
  call  SysReadStringPart
  movlw low StringTable6
  movwf SysStringA
  movlw (high StringTable6) & 127
  movwf SysStringA_H
  call  SysReadStringPart
  movlw low SYSSTRINGPARAM1
  movwf FSR0L
  movlw high SYSSTRINGPARAM1
  movwf FSR0H
  movf  SysStringLength,W
  movwf INDF0
  movlw low SYSSTRINGPARAM1
  movwf SysPRINTDATAHandler
  movlw high SYSSTRINGPARAM1
  movwf SysPRINTDATAHandler_H
  movlw 1
  movwf COMPORT
  call  HSERPRINT256

But it doesn't give me a clue where the compiler overlapped the seg array.

To be clear, the problem does not occur if two separate HSerPrints are done, one for each string, it's only the concatination of the two strings within one print that causes the problem.

[Anobium]

What version of compiler etc. Look at the first line of the ASM

[Jim giordano]

;Program compiled by Great Cow BASIC (0.98.06 2019-06-12 (Windows 32 bit))

[Anobium]

Leave it with me. I need to examine in a debugger.

Thanks for version info.

[Anobium]

Look like a bug.

Check the first code example. This is the easier way to reproduce,

Use the workaround. Shown in second example.

#chip 16f877a

#option explicit

dim seg(9)
dim ii


for ii=1 to 8:seg(ii)=63:next ii  'preset all values to 63

for ii=1 to 8
  epwrite ii,seg(ii)
next ii


passedsub ("abcde"+"fghij")  ' this caused the problem

for ii=1 to 8
  epwrite ii+16,seg(ii)
next ii


sub passedsub( instring as string )

end sub

To work around, concat first then pass as a parameter.

My guess, you know this already.

#chip 16f877a

#option explicit

dim seg(9)
dim ii


for ii=1 to 8:seg(ii)=63:next ii  'preset all values to 63

for ii=1 to 8
  epwrite ii,seg(ii)
next ii

dim mystring as string
mystring = "abcde"+"fghij"
passedsub ( mystring )  ' this will not cause a problem

for ii=1 to 8
  epwrite ii+16,seg(ii)
next ii


sub passedsub( instring as string )

end sub

[Anobium]

Another way to resolve.... add a CONCAT function.

But, we need to resolve to avoid this issue in the future.

#chip 16f877a

#option explicit

dim seg(9)
dim ii


for ii=1 to 8:seg(ii)=63:next ii  'preset all values to 63

for ii=1 to 8
  epwrite ii,seg(ii)
next ii


passedsub ( concat("abcde","fghij") )  ' this resovles the problem

for ii=1 to 8
  epwrite ii+16,seg(ii)
next ii


sub passedsub( instring as string )

end sub

function concat ( instring1 as string, instring2 as string ) as string * 40

  concat = instring1 + instring2

end Function