[OE-core] linux-yocto and linux-libc-headers are again going backwards

[Richard Purdie]

On Thu, 2016-10-06 at 10:07 -0700, Khem Raj wrote:
> > 
> > On Oct 6, 2016, at 2:27 AM, Richard Purdie <richard.purdie at linuxfou
> > ndation.org> wrote:
> > 
> > On Thu, 2016-10-06 at 10:54 +0200, Martin Jansa wrote:
> > > 
> > > Either use 4.7.99+4.8-rc4 like most other recipes do or at least
> > > "4.8~rc4"
> > > which might sort lower than "4.8" (but needs to be verified for
> > > all
> > > packaging backends first).
> > > 
> > > Ignoring all this completely results with what we can see in oe-
> > > core
> > > now:
> > > 
> > > ERROR: linux-libc-headers-4.8-r0 do_packagedata: QA Issue:
> > > Package
> > > version for package linux-libc-headers-dbg went backwards which
> > > would
> > > break package feeds from (0:4.8-rc4-r0.0 to 0:4.8-r0.0)
> > > [version-going-backwards]
> > Sorry, we did screw up there :(. Lots of moving pieces and its hard
> > to
> > catch/remember everything.
> > 
> > I wish we knew for certain whether ~ works properly everywhere. A
> > good
> > case for some unittests I guess…
> I always wished there was a magic wand ( a sanity test) sort of which
> was able to catch the package feed upgradability

There is buildhistory but sadly it catches it too late (after 4.8-rcX
merges and is changed to 4.8, not when 4.7 -> 4.8-rcX).

Cheers,

Richard