Mail Archives: cygwin/2004/09/18/08:51:56
Hello Grzegorz,
Am Freitag, 17. September 2004 um 22:44 schriebst du:
> Hi Gerrit,
> On Fri, 2004-09-17 at 08:42, Gerrit P. Haase wrote:
>> Am Freitag, 17. September 2004 um 02:47 schriebst du:
>> > 2. make libsablevm link against statically compiled libffi.
>>
>> > I am not sure 2) is possible and I don't know how to do it.
>> > If 2) is not possible, would it be possible for gcc maintainer
>> > to include shared version of libffi in gcc-java?
>>
>> You can link against static archives when using 'pass_all' instead of
>> 'file_magic ^x86 archive import|^x86 DLL' to recognise dependent
>> libraries in libtool (change in libtool.m4).
> You mean I should edit /usr/share/libtool/libtool.m4 and change this
> part, that is already specific to cygwin?
> cygwin*)
> # func_win32_libid is a shell function defined in ltmain.sh
> lt_cv_deplibs_check_method='file_magic ^x86 archive import|^x86 DLL'
> lt_cv_file_magic_cmd='func_win32_libid'
> ;;
> It sounds hackish. Terribly hackish. If it's a normal thing you need
> to do at least for some applications under cygwin, why not have some
> kind of a switch for that included into official libtool?
Please submit a patch;)
> Or have I missed something?
It is as is for backward compatibility, new created DLLs don't need
all the stuff with __declspec(import/export), however, some libs still
use it and if you link against one which uses import/export
definitions it may break things when using pass_all. Then there are
still some problems with exporting data, so it may also be needed to
use import/export definitions with DLLs containing data, and then it
would break too. As long as you have libraries which don't use
import/export definitions and since all code is PIC on Windows anyway,
you may safely use pass_all.
Gerrit
--
=^..^=
--
Unsubscribe info: http://cygwin.com/ml/#unsubscribe-simple
Problem reports: http://cygwin.com/problems.html
Documentation: http://cygwin.com/docs.html
FAQ: http://cygwin.com/faq/
- Raw text -