Any change that desyncs things across architectures is active work and effort.
If i386 desktop images are produced, they will be produced exactly the
same way they were produced to date.
Spinning up point releases with or without hwe stack, was not per
architecture so far, but could be modified at point release time.
On 2 February 2016 at 19:03, Bryan Quigley <[email protected]> wrote:
> On Tue, Feb 2, 2016 at 1:28 PM, Dimitri John Ledkov <[email protected]> wrote:
>> could you please drop the HWE enablement stack out of this?
> Hmm.. Let me make it clearer and drop a lot of it.
>> The cost of: testing, validation, release, and bandwidth to mirror it
>> is IMHO large. And costs more than the benefit it will provide.
>> On 2 February 2016 at 17:26, Bryan Quigley <[email protected]> wrote:
>>>> Kernel support is a separate vector. E.g. in Debian it is common to
>>>> install 32-bit userspace with the 64-bit kernel. Thus using all the
>>>> CPU/kernel features, access all the memory, yet have lower memory
>>> Right, but depending on what we decide it will also impact how tested
>>> the HWE stack is on Unity. Say we stop building the x86_32 image
>>> starting with 16.10. Would backporting the x86_32 bit kernel from
>>> 16.10 to 16.04.2 HWE release still happen?
>> Out of scope for this thread.
> I disagree.. .let me reprase.
> If we build Ubuntu 16.04 LTS x86_32 Desktop image will we update it
> with the HWE stack from 16.10/17.04/17.10/18.04 at the point releases
> (16.04.2) like usual?
> I'm not saying that other desktops (Like Kubuntu, and Unity x86_64)
> will still get the HWE like normal. Just will we respin the 32-bit
> Unity CD?
>>>> My argument for dropping .iso, but keeping the packages/archive is as follows:
>>>> * we would like to support upgrades, for those that have 32 bit install
>>>> * but we would like to prevent any new installations
>>> I just want to prevent further bit root for those upgrade users.
>>> There will be even less people testing those now, so I do think we
>>> need a plan to eventually remove Unity from the archive and maybe
>>> migrate those users to another DE? (Unity8 seems to be doing x86_32
>>> releases? the obvious choice for me would be Xubuntu/Mate/Lubuntu but
>>> we don't need where to move today)
>> No need to provide upgrade path. The hardware will simply EOL.
> What do you mean by that exactly?
> Is this with us eventually dropping Unity7 from archive? We still
> need to take some action even if it's just saying.. Sorry Ubuntu
> x86_32 with Unity can't be upgraded any further. But then I don't see
> the harm in saying click here to install Desktop Environment X.
>>>> * because any new installation is amd64 capable, or such is the Ubuntu
>>>> Desktop ISO installer requirement for 16.04 LTS
>>>> * reduce releases.ubuntu.com mirror costs by about a third
>>>> Otherwise, all survey results will remain constant.
>>>> Building images is cheap, however I do not believe we can actually
>>>> adequately support i386 ones for ubuntu desktop:
>>>> * there is no i386-only certified hardware
>>>> * image manual testing has a cost
>>>> * no ubuntu developers use them =)
>>>> Could we start the sunset period with removing flavour dropdown from
>>>> the ubuntu desktop download pages for 16.04? (But keep the i386 images
>>>> on releases.ubuntu.com?)
>>> I'm 100% for that. Still supported (although not certified), but you
>>> really have to know you want to get it.
>>> So maybe a basic plan like:
>>> 1. Announce that Ubuntu 16.04 LTS will be the last officially
>>> supported release of Unity. Keep it on releases.u.c but remove from
>>> main download page. Also announce that x86_32-bit Ubuntu (server
>>> too?) won't be getting HWE?
>> server is out of scope for this discussion, HWE is out of scope. I
>> don't think we ever announce any ubuntu.com website changes. And the
>> website has links to reach to reach the releases.u.c and cdimage.u.c
> I meant to announce in the schedule/release notes the plan for the
> x86_32 release. Last LTS with x86_32 Unity. Make it clear that if
> you go with it, you might not have an upgrade path to the next LTS.
>>> 2. Drop to cdimage for 16.10 with not tested/supported caveat
>>> (continue based on usage numbers)
>>> 3. For 17.04 evaluate migration options and consider dropping Unity7
>>> from x86_32 archive
>> I'd rather say, no action. Reinstall is the best way to "upgrade"
>> those machines to a different ubuntu flavor, or like buy a new
> So is there a point in not just dropping x86_32 Unity from archive in
> 16.10? If we drop Unity7 from the archive we aren't going to produce
> the CD anymore.
> Kind regards,
ubuntu-devel mailing list
Modify settings or unsubscribe at: https://lists.ubuntu.com/mailman/listinfo/ubuntu-devel