3.9-4 VM on cygwin: check that /usr/local/lib/squeak/3.9-4/vm-display-X11.so exists

Klaus D. Witzel klaus.witzel at cobss.com
Mon May 15 23:50:02 UTC 2006

Hi Bert,

on Mon, 15 May 2006 23:57:34 +0200, you wrote:
> "-headless" is an option of the X11 driver, so that's not that  
> surprising.

Right you are. Now I understand why on a debian sans X11 the only option  
was vm-display-null (for running squeak unattended as back-end to apache),  
alternately vm-display-fbdev.

> As for loading the display modules, they have to be shared libraries  
> suitable to be used by dlopen().

This is exactly the point. Will have to find out how dlopen is on cygwin  
(google says that people use LoadLibrary on cygwin, even though dlopen is  

> You only have static libraries, at least that's what I guess from the  
> ".a" extension.

Yes, and I couldn't convince whomsoever to make them non-static (have seen  
the various comments in ltmain.sh, some of which speak about cygwin).

> I guess on Windows that would have to e a DLL.

Yes, if there's need to dynamically load a library the win32 people must  
make a .dll. But nobody here expected that cygwin just uses what is lying  
around (thereby putting the burden on the library dev'er). Porting an app  
to cygwin seems to be easy; doing the same for supportive libs seems to be  
the difference between platform independence and windoze...

> And even if you compile the modules as Windows DLL

This would require changes to the source, beginning with
- http://lists.debian.org/debian-win32/2002/08/msg00001.html
and is not what we want here (have dev'ed some .dll's running on the ISAPI  
interface of MS' web server...).

> you might have to adjust the module loader to look for a ".dll"  
> extension.

I think we better hack the loader for not requesting vm-display-* , looks  
like that would save some time. And/or perhaps vm-display-custom needs to  
be employed for the cygwin job, starting with __declspec(dllimport).

> - Bert -

Your comments (and questions!) where very valuabe to us, thank you very  


More information about the Vm-dev mailing list