[lubuntu-devel] Architecture EOL plans for Lubuntu

[Bryan Quigley]

On Tue, Jul 12, 2016 at 1:27 AM, Walter Lapchynski <wxl at ubuntu.com> wrote:
>
> On Jul 11, 2016 14:11, "Bryan Quigley" <bryan.quigley at canonical.com> wrote:
>> On Mon, Jul 11, 2016 at 3:24 PM, Jan Holtman <oulik.jan at gmail.com> wrote:
>
>> > There are a lot of older very good quality 32 bit machines around,
>> > especially in 3rd world countries.
>> Certainly true, but for many of them the power efficiency is making
>> them less and less a good economic decision.
>
> Unless you consider the fact that these machines are readily available for
> little to no cost. This is true not only in the third world but here in the
> US where public school systems struggle to keep pace with the ever
> increasing system requirements of nearly every operating system out there.
> All the more reason for i386, at least among the Lubuntu camp.

Absolutely and there is a sunk cost (at least work to change) for
already having a machine and there is a capital expense for buying
one.  I was just saying that in some cases the cost of power (which is
hidden on the individual level sometimes) and add up to make some
older machines not worth it.

For example, I had a i386 machine single core with hyper threading
which used about 100 Watts when on.   In this case, my Raspberry Pi 2
was much faster and it only uses 5 Watts when on.

The i386 machine would cost an organization about $40 a year (assuming
8 hours a day, 25 days a month, 16 cents electricity).  Or in other
words, buying the Pi would net a benefit to the organization after a
year and a half ($35 for the Pi and then SD card).

>> The powerpc and alternate cds seem like safer things to consider dropping.
>
> PPC machines are useless without a Linux that supports them and there are
> fewer and fewer. A lightweight distro like Lubuntu makes logical sense in
> this case.
>
> As for alternate, it's a nice alternative to netboot which requires a good
> Internet connection. In many parts of the world, that's asking too much.
> Similarly, ubiquity has greater system requirements than Lubuntu, so it
> doesn't make a lot of sense. If there was a non-graphical frontend to
> debian-installer, that would be great because then server could adopt it
> too.
>
> That being said, I know that we can't do everything. At some point, we have
> to continue to refine our focus, lest we do everything but nothing well. And
> as we continue to grow support for mobile and IoT, our needs change, so I
> can see the need to remove some extraneous pieces. However, for an
> increasing segment of this market, x86 is a norm.
>
> That being said, I can envision reasons to get rid of the alternate and the
> PPC images, but maybe removing i386 is a bit short-sighted.

Indeed, I just wanted to explicitly have the conversation with every
flavor.  Based on every result* I've seen, I'd recommend Lubuntu be
the last (desktop) flavor to drop i386 ~ whenever that happens.

For alternate cds: I'm surprised they're still needed, given that I
can run Lubuntu (i386) livecd well with just 256 MB of ram.  Having
said that it doesn't look like the costs of building alternates are
very high.

PPC: This was mentioned in passing that this arch might be worth
considering for removal in the future (actually easier to rm then
i386) and seeing as Lubuntu is one of only two desktop flavors that
provide I'd figure I'd include it in this discussion.   If you have a
guesstimate of what the effect of dropping PPC for 16.10 would be for
Lubuntu that would be very useful (both negative - current X users
lose support after 3 years, and positive - better focus, PPC specific
support issues?) to start having a wider discussion about PPC.

Kind regards
Bryan

*Survey results, torrent results, and being the lowest desktop memory user.