Hi Jürgen:
Well that trivial program confirms my hypothesis which is that I am the first 
person to try using libapl on macOS.
 I have been doing similar things to your program using both the command line 
and also within Xcode (Apple’s GUI Developer tool). From my experiments I had 
concluded that libapl.so, 
which is a bundle, will not work on macOS. 

We need a dylib. What I have not yet fully figured out is how to change things 
for libtool so as to fix things.
Not sure if I have to change the compile args or just the linker ones.
My understanding is that the .o files are PIC and it is the linker which builds 
them into an image that's relocatable.

For dynamic libs the linker jus puts a reference in the exec file to the 
library image but does not include it’s content in the exec.
The dynamic loader will take care of that and also resolving addresses at 
launch time.

Gandalf:~ pteeson$ cd /Volumes/Data/Development/apl-1053\ after\ Make/src 
Gandalf:src pteeson$ cat libapl_test.c
#include "libapl.h"

main(int argc, char * argv[])
    init_libapl(argv[0], 0);
    return apl_exec("⎕←2 3 ⍴⍳6");
Gandalf:src pteeson$ gcc -o libapl_test libapl_test.c -L .libs -lapl
ld: can't link with bundle (MH_BUNDLE) only dylibs (MH_DYLIB) file 
'.libs/libapl.so' for architecture x86_64
clang: error: unable to execute command: Segmentation fault: 11
clang: error: linker command failed due to signal (use -v to see invocation)

I also ran the command with the -v option and this is what was produced:
I understand everything below. Just don’t know how to fix libtool for macOS

This is the key line with the args to ld:
 -demangle -dynamic -arch x86_64 -macosx_version_min 10.10.0 -o libapl_test 

Gandalf:src pteeson$ gcc -v -o libapl_test libapl_test.c -L .libs -lapl
Apple LLVM version 6.1.0 (clang-602.0.53) (based on LLVM 3.6.0svn)
Target: x86_64-apple-darwin14.5.0
Thread model: posix
 -cc1 -triple x86_64-apple-macosx10.10.0 -emit-obj -mrelax-all -disable-free 
-disable-llvm-verifier -main-file-name libapl_test.c -mrelocation-model pic 
-pic-level 2 -mdisable-fp-elim -masm-verbose -munwind-tables -target-cpu core2 
-target-linker-version 242.2 -v -dwarf-column-info -resource-dir 
 -fdebug-compilation-dir /Volumes/Data/Development/apl-1053 after Make/src 
-ferror-limit 19 -fmessage-length 139 -stack-protector 1 -mstackrealign 
-fblocks -fobjc-runtime=macosx-10.10.0 -fencode-extended-block-signature 
-fmax-type-align=16 -fdiagnostics-show-option -fcolor-diagnostics -o 
/var/folders/w1/081y692x5x30c39mcr8lxwkh0000gn/T/libapl_test-2b7dfd.o -x c 
clang -cc1 version 6.1.0 based upon LLVM 3.6.0svn default target 
#include "..." search starts here:
#include <...> search starts here:
 /System/Library/Frameworks (framework directory)
 /Library/Frameworks (framework directory)
End of search list.
 -demangle -dynamic -arch x86_64 -macosx_version_min 10.10.0 -o libapl_test 
-L.libs /var/folders/w1/081y692x5x30c39mcr8lxwkh0000gn/T/libapl_test-2b7dfd.o 
-lapl -lSystem 
ld: can't link with bundle (MH_BUNDLE) only dylibs (MH_DYLIB) file 
'.libs/libapl.so' for architecture x86_64
clang: error: linker command failed with exit code 1 (use -v to see invocation)

> On Jun 13, 2018, at 1:58 PM, Juergen Sauermann 
> <juergen.sauerm...@t-online.de> wrote:
> Hi Peter,
> a simple C program, say libapl_test.c  would be this:
> #include "libapl.h"
> int
> main(int argc, char * argv[])
> {
>    init_libapl(argv[0], 0);
>    return apl_exec("⎕←2 3⍴⍳6");
> }
> Compile it in src like so:
> gcc -o libapl_test libapl_test.c -L .libs -lapl
> Executing it gives:
> eedjsa@server66:~/projects/juergen/apl-1.7/src$ 
> <mailto:eedjsa@server66:~/projects/juergen/apl-1.7/src$> ./libapl_test
> 1 2 3
> 4 5 6
> /// Jürgen
> On 06/13/2018 05:34 PM, Peter Teeson wrote:
>> Hi Dirk:
>> Please tell me what OS you were using? 
>> thanks….
>> Peter
>>> On May 20, 2018, at 4:59 PM, Dirk Laurie <dirk.lau...@gmail.com> 
>>> <mailto:dirk.lau...@gmail.com> wrote:
>>> I did this three years ago, using SVN 570 of GNU APL. In an ideal
>>> world, I would have checked after every SVN update that my application
>>> still works. In the real world, I have not touched it since and cannot
>>> remember much. :-(
>>> I currently have SVN 1048. When I tried it my application [1] (which
>>> runs GNU APL in parallel with Lua) just now, the Lua 5.2 version that
>>> I made on 29 May 2015 still works in a simple test.
>>> $ lua -l gnuapl
>>> Lua 5.3.4  Copyright (C) 1994-2017 Lua.org, PUC-Rio
>>> …/gnuapl$ lua5.2 -l gnuapl
>>> Lua 5.2.4  Copyright (C) 1994-2015 Lua.org, PUC-Rio
>>>> =gnuapl.exec"4 4⍴⍳16"
>>> 1  2  3  4
>>> 5  6  7  8
>>> 9 10 11 12
>>> 13 14 15 16
>>> This seems to confirm that there is nothing wrong with libapl.so.
>>> Unfortunately I have no simple C main program, since everything runs
>>> through Lua. In particular, Lua's 'package.loadlib' function is used
>>> to load the current libapl.so. The code for that function is way above
>>> my code-reading ability.
>>> Sorry that I can't offer more help.
>>> -- Dirk
>>> [1] Those that are reasonably fluent in Lua and its C API can try it
>>> out: https://github.com/dlaurie/lua-gnuapl 
>>> <https://github.com/dlaurie/lua-gnuapl>

Reply via email to