Re: [Opensync-users] Problem compiling synce plugin in FC4

[Armin B. <arm...@de...>]

R=C3=A9my Amouroux wrote:
> Waiting for an answer from the synce-devel maintainer, I wanted to go
> further compiling the synce plugin.
>=20
> I came back to what is in CVS by redoing a checkout and created a rra
> directory containing all the include files provided by
> synce-devel-0.9.1-3.fc4.
>=20
> I'm using the last gcc provided with the FC4: gcc-4.0.1-4.fc4
>=20
> So, now, configure works like a charm and I have no include errors sinc=
e
> rra/ is there.
>=20
> But, I have compilation errors :-(
>=20
> Here is the traces.
>=20
> gcc -DHAVE_CONFIG_H -I. -I. -I..
> -I/home/amourouxr/Tools/usr/local/include/rra
> -I/home/amourouxr/Tools/usr/local/include -Wall -Werror
> -I/usr/include/glib-2.0 -I/usr/lib/glib-2.0/include
> -I/usr/include/libxml2
> -I/home/amourouxr/Tools/usr/local/include/opensync-1.0 -g -O2 -MT
> synce_plugin.lo -MD -MP -MF .deps/synce_plugin.Tpo -c synce_plugin.c
> -fPIC -DPIC -o .libs/synce_plugin.o
> cc1: warnings being treated as errors
> synce_plugin.c: In function 'm_report_todo_changes':
> synce_plugin.c:256: warning: pointer targets in passing argument 1 of
> 'osync_print_binary' differ in signedness
> synce_plugin.c: In function 'm_report_cal_changes':
> synce_plugin.c:304: warning: pointer targets in passing argument 1 of
> 'osync_print_binary' differ in signedness
>=20
> After removing the two lines (calls to osync_print_binary seems to be
> only for debugging purposes), everything compiles fine.
>=20

Thanks. i committed a fix for this.


> I'll try synchronisation tomorrow.
>=20
> one comment below about your answer, armin.
>=20
>=20
>>I think the correct way is to do it like this:
>>
>>rra and synce get installed into a directory ${includedir}/rra-$version=
/rra
>>
>>this way its possible to use the rra/ prefix in the #include statement
>>(which works also a sort of a namespace so includes dont overlap) but i=
t
>>is also possible to install several versions separately.
>=20
>=20
> I'm not sure it's really interesting to have the version in the
> directory name. It makes things difficult for the developper using the
> library since it has to provide configuration files managing the versio=
n
> information and cannot rely on a default location.
> If the maintainer of the package really want to allow several
> installations of the same package, he has to make that package
> relocatable. Thus, with a simple --with-rra=3DDIR, you can use differen=
t
> version of the library.

ok, right. with opensync i use pkgconfig so i always get the correct
include path from it but i can also install another version and
overwrite the include.

>=20
> Regards
>=20
> RemyA