Weird output when printing the result of a comparison

[Roland Chastain]

Hello gentlemen!

I have this program (a test program for a library).

Code: Select all

dim as any ptr library = dylibload("mater")

if library = 0 then
  print "Failed to load the dynamic library, aborting program..."
  end 1
end if

dim SolveMate as function ( _
  byval as const zstring ptr, _
  byval as const integer, _
  byval as const integer _
) as zstring ptr

SolveMate = dylibsymbol(library, "SolveMate")

if SolveMate = 0 then
  print "Could not retrieve the function's address, aborting program..."
  end 1
end if

const SEARCH_ALL_MOVES = 1
const SEARCH_CHECK_ONLY = 0

print("b7a8n" = *SolveMate("b7/PP6/8/8/7K/6B1/6N1/4R1bk w KQkq -", 3, SEARCH_ALL_MOVES))
print("h6c1" = *SolveMate("8/8/1p5B/4p3/1p2k1P1/1P3n2/P4PB1/K2R4 w KQkq -", 3, SEARCH_ALL_MOVES))
print("d6c7" = *SolveMate("2N5/8/k2K4/8/p1PB4/P7/8/8 w KQkq -", 4, SEARCH_ALL_MOVES))
print("g5g2" = *SolveMate("rnbK2R1/p6p/p1kNpN1r/P3B1Q1/3P1p1p/5p2/5p1b/8 w KQkq -", 4, SEARCH_ALL_MOVES))
print("h7g7" = *SolveMate("8/1n2P2K/3p2p1/2p3pk/6pr/4ppr1/6p1/1b6 w KQkq -", 3, SEARCH_ALL_MOVES))
print("b7b8r" = *SolveMate("4K1R1/PP2P3/2k5/3pP3/3B4/6P1/8/8 w KQkq -", 3, SEARCH_ALL_MOVES))
print("e7e8b" = *SolveMate("8/2P1P1P1/3PkP2/8/4K3/8/8/8 w Qkq -", 3, SEARCH_ALL_MOVES))
print("h3g5" = *SolveMate("3nn3/2p2p1k/1p1pp1p1/p2B3p/r2B2N1/7N/8/7K w KQkq -", 12, SEARCH_CHECK_ONLY))

dylibfree(library)

Here is the output of the program under Linux:

Code: Select all

-1
-1
-1
-1
-1
-1
-1
-1

And here is the output under Windows:

Code: Select all

-1
              -1
              -1
-1
-1
              -1
              -1
              -1

Hex dump:

00000000 2D 31 0D 0A 20 20 20 20 20 20 20 20 20 20 20 20 -1..............
00000010 20 20 2D 31 0D 0A 20 20 20 20 20 20 20 20 20 20 ..-1............
00000020 20 20 20 20 2D 31 0D 0A 2D 31 0D 0A 2D 31 0D 0A ....-1..-1..-1..
00000030 20 20 20 20 20 20 20 20 20 20 20 20 20 20 2D 31 ..............-1
00000040 0D 0A 20 20 20 20 20 20 20 20 20 20 20 20 20 20 ................
00000050 2D 31 0D 0A 20 20 20 20 20 20 20 20 20 20 20 20 -1..............
00000060 20 20 2D 31 0D 0A ..-1..

Any idea of what happens? Why those extra spaces?

I tried with FBC 1.08 and FBC 1.09 (both for 32-bit): same result.

FYI, the program comes from this project: https://gitlab.com/rchastain/mater-library

[SARG]

Check if in SolveMate or in a nested function there would be a print.

[srvaldez]

Hello Roland Chastain
can't build the library in windows, what version of freePascal are you using ?
or is it Delphi ?

[dodicat]

Hello roland.
Win 10

Code: Select all

-1
-1
-1
-1
-1
-1
-1
-1
Press any key to continue . . . 

srvaldez
you need
{$mode delphi}
in mater.pas, materutils.pas, matercore.pas.
matertypes.pas is OK without it.
(I use Geany ide for freepascal, so I need first principles and not any preset Lazarus flags).
( {$mode delphi} only applies in the unit it is used in. )

SORRY Roland, I just noticed 32 bits, I tested 64 bits.
I will test 32 bits later.

[Roland Chastain]

Thank you all for your answers.

can't build the library in windows, what version of freePascal are you using ?

I compile with these options:

Code: Select all

fpc -Mobjfpc -Sh mater.pas

You can also use the 'Delphi' mode as dodicat suggested:

Code: Select all

fpc -Mdelphi mater.pas

Otherwise, I use FPC 3.2.0, but I don't think it matters.

I just noticed 32 bits, I tested 64 bits.

Good to know.

[Roland Chastain]

Check if in SolveMate or in a nested function there would be a print.

Yes, it could be an explanation, but I don't think there is a print. The problems occurs only with this demo program, not with other.

[SARG]

Another idea :
put the pointed value in a string mystrg=*SolveMate(.....)
then
print "xx";mystrg;"xx"
print ("b7a8n" = mystrg)

[srvaldez]

@dodicat, thank you
@Roland Chastain, it works ok if using the gcc back-end but fails as you described when using the asm back-end, tested on 32-bit version

[Roland Chastain]

@Roland Chastain, it works ok if using the gcc back-end but fails as you described when using the asm back-end, tested on 32-bit version

Thank you for the information. But I don't know how to choose the back-end. Should I reinstall the compiler from another package, or is it a command line option?

P.-S. I used this package. Should I try use this one?

[dodicat]

I tested 32 bits now also.
Same problem.
I made LVerbose=True, the -1 then are not spaced.
I set it back to FALSE

puts str(...) also gives weird results in gas
example
...
puts str("e7e8b" = *SolveMate("8/2P1P1P1/3PkP2/8/4K3/8/8/8 w Qkq -", 3, SEARCH_ALL_MOVES))
...

Code: Select all

?‼³
¸'³
ð7³
 8³
P8³
?8³
°8³
à8³ 

puts is OK -gen gcc 32 bits

with fb 1.09
put #cmdline "-gen gcc"
at the top of the code.

[srvaldez]

Roland Chastain, that looks like the latest official release
since version 1.09.0 you can put compiler options in the source as dodicat mentioned, put something like this at the top of your source #cmdline "-gen gcc"

[Roland Chastain]

Very interesting informations. Thank you dodicat.

with fb 1.09
put #cmdline "-gen gcc"
at the top of the code.

I will try tomorrow. (Now, I am at home when there is only Linux.)

[Roland Chastain]

since version 1.09.0 you can put compiler options in the source as dodicat mentioned, put something like this at the top of your source #cmdline "-gen gcc"

OK, I didn't know. Great, thank you!

[SARG]

I tried to reproduce the problem with this code but unsuccessfully :

Code: Select all

function  solve (z as zstring ptr,a as integer,b as integer) as zstring ptr
	static as string zstr
	zstr="b7a8n"
	'print *z
	return strptr(zstr)
End Function

dim SolveMate as function (z as zstring ptr,a as integer,b as integer) as zstring ptr
SolveMate=@solve()

print *SolveMate("aaa",12,15)
print ("b7a8n" = *SolveMate("aaa",12,15))
print ("b7a8n" = *SolveMate("aaa",12,15))
print ("b7a8n" = *SolveMate("aaa",12,15))
print ("b7a8n" = *SolveMate("aaa",12,15))
print ("b7a8n" = *SolveMate("aaa",12,15))
print ("b7a8n" = *SolveMate("aaa",12,15))

sleep

Try on your side.
And if you can please upload the dll and I'll test.

[dodicat]

Here is the 32 bit pascal dll
https://www.mediafire.com/file/f3a294yz ... r.zip/file
I am also trying to simulate the error, as yet no luck.