PRINT FORMAT anomality

[rikky]

Hello

I'm using the newest bacon, just released.

FUNCTION FORMAT$(digits)
 
 'zero_digits$ = "%.0f"
 'one_digit$ = "%.1f"
 'two_digits$ = "%.2f"
 
 LOCAL return$
 
 return$ = "%." & STR$(digits) & "f"
 
 RETURN return$
 
 'PRINT VAL("123") / VAL("8") FORMAT PRINT_FORMAT$(3)
 
END FUNCTION

bla$ = FORMAT$(2)

PRINT VAL("1.23456789") FORMAT "%.2f"
PRINT

PRINT VAL("1.23456789") FORMAT bla$
PRINT

PRINT VAL("1.23456789") FORMAT FORMAT$(2)
PRINT

PRINT VAL("1.23456789") FORMAT "%.2f" TO blabla$
PRINT blabla$

PRINT VAL("1.23456789") FORMAT bla$ TO blabla$
PRINT blabla$

PRINT VAL("1.23456789") FORMAT FORMAT$(2) TO blabla$
PRINT blabla$

result:

1.23
1.23
1.23
1.23
1.23
*** Error in `./test': free(): invalid pointer: 0x01e79719 ***
ERROR: signal ABORT received - internal error. Try to compile the program with TRAP LOCAL to find the cause.
I'm sure that it did work, not too long ago.

   Rik.

[bigbass]

Hello Rik

this may or not be a bug (that's not my decision)
I just offer solutions to get code
to compile and run

the FUNCTION with a return type char*



FUNCTION FORMAT$(digits) TYPE char*

worked for me


hope that could be useful
Joe

[vovchik]

Dear Rik and Joe,

Thanks for pointing out the anomaly (Rik) and for the fix (Joe). TYPE char* works fine...

And, as I just discovered, so does STRING:

FUNCTION FORMAT$(digits) TYPE STRING

With kind regards,
vovchik

[Pjot]

Hi Rik,

This is a complicated situation, and I need more time to find a good solution. For the record, the code does not work in previous versions of BaCon either. It cannot be compiled even before 4.3. So it never would work in BaCon.

If we look at the cause of the problem, you'll see why it never would have worked. Namely, the reason is that both the PRINT..FORMAT..TO statement and also the function 'FORMAT$' share the same internal variable, and both try to free them.

Exactly this is the reason that I started to cleanup the code in 4.2 and 4.3, to get rid of as much of these internal variables. Because they can lead to a clash.

The solution of vovchik and Joe is a valid workaround for this particular situation. I recommend to apply it for now.

But the root cause seems to be more severe, and this is why I need more time to find a long-term fix.

Best regards,
Peter

[rikky]

I already had another workaround, so there is no need to rush.

bla$ = FORMAT$(2)
PRINT VAL("1.23456789") FORMAT bla$ TO blabla$
PRINT blabla$
I was pretty sure it was working so I tried some museum versions, and

BaCon version 4.1.1 on Linux armv7l - (c) Peter van Eerten - MIT License.
works.

result 4.1.1 :

1.23
1.23
1.23
1.23
1.23
1.23
 Rik.

[Pjot]

Well, thats weird, because for me it does not even compile:

$ bash current/bacon-4.1.1/bacon.sh -v

BaCon version 4.1.1 on Linux x86_64 - Shell - (c) Peter van Eerten - MIT License.

$ bash current/bacon-4.1.1/bacon.sh -y -p -o -g -@ debug1
Converting 'debug1.bac'... 37
char*:MAIN:__b2c__loop_result__b2c__string_var char*:MAIN:blabla__b2c__string_var
long:FORMAT__b2c__string_var:digits char*:FORMAT__b2c__string_var:return__b2c__string_var
Converting 'debug1.bac'... done, 37 lines were processed in 1 seconds.
Analyzing dependencies... concat error malloc memcheck minmax str utf8  
Applying indentation... done.
Compiling 'debug1.bac'... cc  -g -c debug1.bac.c
cc -o debug1 debug1.bac.o    -lm

debug1.bac.c:74:37: warning: implicit declaration of function '__b2c__string_var'; did you mean 'CL__b2c__string_var'? [-Wimplicit-function-declaration]

INFO: see full error report (y/[n])?


So there seems to be another issue in 4.1.1 as well But I will fix it for the latest only, of course.

BR
Peter

[rikky]

bacon.4.1.1.sh also doesn't compile over here.
It's the executable that does it.

And to be honest,
I did NOT take this executable from the website, but from my own system somewhere.

So it might be a bacon.4.1.1 FOSSIL.

[Pjot]

Hi rikky,

The issue now is fixed in the latest beta.

Best regards
Peter

[rikky]

Yes,
Now it works

result:

1.23
1.23
1.23
1.23
1.23
1.23
Thanks. 

 Rik.