Re: Travis and OCaml, and even ARM

[Anil Madhavapeddy <anil@xxxxxxxxxx>]

We mentioned on the call today that Travis might not be able to build the Xen 
backend since it's an i386 VM, but it does work fine (since we don't actually 
use any system libraries in the build)

https://travis-ci.org/mirage/mirage-www/jobs/11897190

So we can use Travis for a publishing platform too...

-anil

On 30 Sep 2013, at 17:04, Anil Madhavapeddy <anil@xxxxxxxxxx> wrote:

> Good news everyone! I just had an exciting build failure that I wanted to 
> share with you.
> 
> Dave and I have been switching a number of our OCaml GitHub repositories to 
> use Travis continuous integration testing.  This is good enough for any tests 
> that can run in an Ubuntu/i386 VM without hosting any infrastructure 
> ourselves.  Details here:
> http://anil.recoil.org/2013/09/30/travis-and-ocaml.html
> 
> If you look at the bottom there, I used Ian Campbell's foreign-chroot 
> instructions to get OCaml/ARM working too.  I've been activating this on 
> various Mirage repositories to test our builds on ARM without having to have 
> a box available, and so this is making it much easier to reproduce bugs such 
> as this shared-memory-ring failure:
> https://github.com/djs55/shared-memory-ring/issues/6
> ...just via this Travis log:
> https://travis-ci.org/mirage/mirage-platform/jobs/11964317
> 
> Now we just have to fix it, but that's pretty easy right? :-)
> 
> Are there any brave souls out there that might try to figure out how to get a 
> Xen environment working from within Travis?  It would only need to boot 
> Mirage kernels, so it's ok to be a bit slow...
> 
> (I expect to finish the conversion of all the Mirage release repositories to 
> Travis this week, and it would be great to see XAPI repos like xenopsd also 
> activated, which I believe Dave is working on).
> 
> -anil