Hi,
Thanks
Stefan
--
You received this message because you are subscribed to the Google Groups "HORN Development" group.
To post to this group, send email to horn-dev...@googlegroups.com.
To unsubscribe from this group, send email to horn-developme...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/horn-development?hl=en.
Please correct me if I'm wrong, but I think OpenWrap and Horn target
different problems.
-Michael
Sent from my iPhone
Building trunks against trunks against trunks is entirely possible using openwrap, you just need to go through a package / deployment phase in between.
Build nhib trunk produces nhibernate-3.1.0[trunk].wrap
Then pull activerecord trunk, compile it against the new package.
You can do all that by keeping all your local builds in an OpenWrap repository that you can set wherever you want on your machine.
OpenWrap supports triggering builds and compiles packages out of the text output that those builds trigger, making a lot of the work that existed in descriptors redundant. Horn could be turned into a tool that can pull source repositories, replace packages where appropriate and rebuild against the new versions if needed. The only thing you need is to generate a descriptor and let the system do the resolve for you, provided the projects can be made to use openwrap.
As it's pretty trivial, I think it's much more productive to maintain a fork for projects that require it and keep the descriptor and changes to project files than it is to invest time in the source-only solution that existed in horn.
That said, there's a feature we're adding to openwrap (mainly so we can auto-publish to symbol servers), and that's including sources in a package, so if you wanted to do source-only dependency tree resolving, including compilation, there's only a step there to integrate with the existing code.
From my perspective, and I suppose I would say that, package management is going to be a two-horse race in the coming months, and I very much doubt continuing with horn as it operated before is going to be efficient or time well invested. There's a featureset that would find a natural home in OpenWrap, and I'm happy helping out and integrating horn features in OpenWrap, I think it'd be more efficient, provide a better competitive advantage considering the synergies.
Then again, that may just be me, I don't want to prevent anyone from resurrecting hornget if they think that's where the future is.