Re: "load" on Windows

[Gisle Vanem]

"Eli Zaretskii" <address@hidden> wrote:

> I don't understand the problem you are describing.  First, when Make
> is built, it produces gnumake.exe, not make.exe.  And second, if
> make.exe that runs is on PATH, rather than in the current directory,
> it's a different executable, and there should be no problem linking
> it.

I simply rewrote the build process to generate make.exe instead of
gnumake.exe. But this is a minor issue. The "load" is what I want to
resolve.

> > My little mk_test.dll is simply trying to echo back some text, but my
> > function is never called.  I checked with "pedump mk_test.dll" that the 
> > symbols "mk_test_gmk_setup" and "plugin_is_GPL_compatible" are exported.
>
> How did you build your DLL?  In particular, did you link it against
> libgnumake-1.dll.a, the import library produced by the build process?

Yes, I modeled this after build_w32.bat. But I made this into a makefile 
with:

make.exe: $(OBJECTS) libsubproc.a
 $(CC) -o $@ $^ -Wl,--out-implib=libgnumake-1.dll.a -luser32 -ladvapi32

And the mk_test.dll:
  $(CC) $(CFLAGS) -shared mk_test.c -o mk_test.dll ./libgnumake-1.dll.a

The mk_test.dll looks okay and 'gnumake -d test_dll' produces no errors
like the one from load.c. That means the dll is okay AFAICS. And make.exe
works fine too. Except for the "load" statement.

Since I do get "Loaded modules: mk_test.dll" it mean it's loaded okay. But
it's never run. So it must be some issue with how I use the extension. I looked 
at section 12.2.4 in the docs and cooked it from that. 


--gv