Re: ipkg dependency problems

Hello Neil,



Tuesday, October 3, 2006, 3:02:06 PM, you wrote:

> I'm trying to test gpe-expenses on my iPAQ (h3900) prior to the first
> release. I'm unable to build anything in the familiar-build using  
> bitbake and OE on amd64 and I am also unable to build the entire  
> gpe-image on i386 using OE itself. I don't have sufficient disc space
> on i386 to build both the OE and familiar-build trees.

> I am only able to build certain components of GPE via OE on i386. I've
> created working .bb files for the necessary packages and the packages
> build OK when called directly.

[]

> In theory, there is no reason why libqof1 and gpe-expenses would not  
> build with the current Familiar 0.84 library versions of libglib-2.0  
> and libsqlite0 - I've tried to ensure this during development.

  Neil, so what exactly DISTRO do you use to build against
org.openembedded.dev? I hope, that's "familiar-unstable"? I myself
interested to have a configuration for OE.dev to build
fam0.8.4-compatible packages, and am working on that. So far I have
some package built ok, and at least glibc used is the same version as
used by fam0.8.4. But I didn't yet have chance to try to install the
package on device, and I suspected that it may drag other
dependencies, like GTK+, and you report confirms this.

  Well, what I can say is that there're people working on this
(myself, maybe someone else too), so this will be fixed soon. I don't
have exact ETA though, but as that's active task on my list, I hope to
have some result within 2 weeks.

> Any comments on the .bb files would be welcome. CVS access available  
> via the homepages given in the relevant .bb snippet if anyone wants to
> test locally. Neither package is ready for release just yet.

  Thanks for working on expenses app for GPE! Actually, response to
your previous mail regarding it is still on my todo ;-I, but as I'm
neck-deep in low-level stuff like kernels and buildsystems, I just
don't want to start discussion about things I don't have currently
time for ;-(. Anyway, I'd be glad to test it as time permits.


-- 
Best regards,
 Paul                            mailto:[email protected]