REPLACE$ & ALIGN$ bugs (SOLVED)

[rikky]

Hello

Since the bug forum is occupied I am reporting here.

I have this program:

PRINT
PRINT VERSION$

string1$ = "bla " & NL$ & "blabla bla bla" & NL$ & "blabla bla bla" & NL$ & "blabla bla bla" & NL$ & "blabla bla bla" & NL$ & "blabla bla bla" & NL$ & "blabla bla bla" & NL$ & "blabla bla bla" & NL$ & "blabla bla bla"

PRINT
PRINT "string1:"
PRINT string1$

string2$ = REPLACE$(string1$,NL$,"")
PRINT
PRINT "string2$ = REPLACE$(string1$,NL$,'')"
PRINT string2$

string2$ = REPLACE$(string2$," ","")
PRINT
PRINT "string2$ = REPLACE$(string2$,' ','')"
PRINT string2$

string3$ = REPLACE$(REPLACE$(string1$,NL$,"")," ","")
PRINT
PRINT "REPLACE$(REPLACE$(string1$,NL$,''),' ','')"
PRINT "=>" & string3$ & "<= huh? it's gone."

'more fun

PRINT
PRINT "ALIGN$(string1$,64,0)"
PRINT ALIGN$(string1$,64,0)

PRINT
PRINT "ALIGN$(string2$,64,0)"
PRINT ALIGN$(string2$,64,0)
result:

4.8.1

string1:
bla
blabla bla bla
blabla bla bla
blabla bla bla
blabla bla bla
blabla bla bla
blabla bla bla
blabla bla bla
blabla bla bla

string2$ = REPLACE$(string1$,NL$,'')
bla blabla bla blablabla bla blablabla bla blablabla bla blablabla bla blablabla bla blablabla bla blablabla bla bla

string2$ = REPLACE$(string2$,' ','')
blablablablablablablablablablablablablablablablablablablablablablablablablablablablablablablablabla

REPLACE$(REPLACE$(string1$,NL$,''),' ','')
=><= huh? it's gone.

ALIGN$(string1$,64,0)
bla blabla bla bla blabla bla bla blabla bla bla blabla bla bla
blabla bla bla blabla bla bla blabla bla bla blabla bla bla    

ALIGN$(string2$,64,0)
ERROR: signal for SEGMENTATION FAULT received - memory invalid or array out of bounds? Try to compile the program with TRAP LOCAL to find the cause.
double free or corruption (!prev)
ERROR: signal ABORT received - internal error. Try to compile the program with TRAP LOCAL to find the cause.


The version 4.8.1 is a fossil from Thursday 18/04/2024 08:48:16.

The issues:
- REPLACE$(REPLACE$(string1$,NL$,''),' ','')  => nix
- ALIGN$(string2$,64,0)   =>  signal for SEGMENTATION FAULT.

 Rik.

[Pjot]

Hi rik,

Thanks for your report on 4.8.1 beta!

Your report contains two problems.

1. An issue with nested string functions - this is a regression caused by an optimization in the internal tokenizer. It omitted the check for string variables, causing the detection of nested string functions to fail. Interestingly, my regression tests never failed, and neither did the BaCon version of BaCon ...so I have designed a new test for this particular situation to detect similar problems in the future.
2. An issue with ALIGN$ which only occurs when (1) the target string does not contain spaces and (2) the target string does not contain newlines and (3) the target string is not UTF8 and (4) the requested string width is smaller than the string itself. No idea how you came up with this rare edge case, but it should work nevertheless, and not cause a segmentation fault! It turned out that there was a 1-byte mismatch in a memory write, which is fixed now.

Please download the updated code here or use 'fossil update' in your source tree.

Your reports are very valuable, please do not hesitate to report other problems!

Thanks again,

Peter