Re: Bug#760114: transition: kfreebsd-kernel-headers
Control: tags -1 confirmed
On 10/09/14 22:10, Steven Chamberlain wrote:
> Hi Emilio,
>
> On 10/09/14 20:16, Emilio Pozuelo Monfort wrote:
>> [...] what
>> packages are involved, what packages need rebuilds, and of those, which ones
>> currently fail.
>
> The root of this is kfreebsd-source-10.0, from the kfreebsd-10 source
> package.
>
> kfreebsd-kernel-headers [kfreebsd-any] has Build-Depends on that exact
> version, so it needs a new upload to use kfreebsd-source-10.1. That has
> been done in experimental.
>
> I don't think anything *requires* a rebuild for this, but
> kfreebsd-kernel-headers is an indirect dependency of build-essential on
> kfreebsd. I expect you'll want us to check we didn't break anything.
> But I was hoping for some hints on how far to go with this.
>
> If something does break, it would be on kfreebsd only, and something
> porters then need to fix.
>
> A simple 'apt-cache rdepends' points to freebsd-glue, freebsd-libs (both
> have been rebuilt against new kfreebsd-kernel-headers in experimental),
> and glibc.
>
> Some other packages set an explicit Build-Depends on
> kfreebsd-kernel-headers, although they don't really need to:
>
> freebsd-smbfs
> freebsd-utils
> gcc-4.8
> gcc-4.9
> gnat
> gnome-mplayer
> pd-iemambi
> pmacct
> sash
> ufsutils
> wine
> zfsutils
>
> I haven't test-rebuilt these yet, but would glibc and everything from
> that list be confidence enough that the 10.0 -> 10.1 changes didn't
> obviously break something?
>
> (AFAIK the 9.0 -> 9.2 -> 10.0 changes in the past year were done without
> much/any co-ordination at all with the release team, but I'm trying to
> do the right thing by asking here first).
Yes, this doesn't seem like a "traditional" transition. So as long as you think
there won't be any/much breakage, and you fix the potential fallout, I think you
can go ahead with this. Of course doing the test rebuilds *before* starting this
would be a good idea to make sure things will still work fine.
Emilio
Reply to: