On Fri, May 06, 2016 at 06:04:32AM +0000, Noel Torres wrote:
>
> Vince Mulhollon <vince@???> escribió:
>
> >It'll hit the embedded world pretty hard.
> >
> >Today you can buy a brand new soekris box that only runs i586. Brand
> >new off the shelf, today. My 6 or 7 year old one is running right now
>
> Very valid points. However, we need to pick our battles. At this
> moment, we have very scarce manpower, and a prime objective: get rid
> of systemd for Jessie (it being usable) and Ascii (completely).

It isn't even likely to impact Jessie, since the problems are only now
being introduced into sid. THe responses I see in the bug reports
suggest to me that Debian will not consider fixing the problem, and is
dropping 586 altogether.

The question for us, when we start to get serious about ascii, is
whether we consider this a reasonable approach for our users
although some of whom seem to still have use for a 586, or whether is
the same kind of attitude that leads to WONTFIX for numerous systemd
problems.

Of course we don't need any kind of decision ontil we get serious
about ascii, but airing the issue beforehand might inform our decision
when we finally do get around to it.

-- hendrik

> Adam
> Borowski expressed it way better than I could:
>
> ==8<==================
> Adam Borowski <kilobyte@???> escribió:
> >Reverting this in a derivative is possible, although it lands you pretty
> >much exactly in Raspbian's position.
> >
> >You'd need to:
> >* reconfigure and rebuild kernel for -585 flavour
> >* undo the not-yet-done merging of libc6-i686
> >* (no source changes) rebuild every package!
> >* watch out for regressions

The main thing I see is that we'd have to use the appropriate flags for
gcc to generate 586 code instead of 686 code.

But yes, it would involve rebuilding every package.

-- hendrik

> ==8<==================
>
> After that (or even for Ascii), we can broaden our view of "user
> freedom to choose" to other chosings, and i585 would be a good first
> step, but at this moment I (humbly) think it is not worth (even
> being very important).

At this moment, I agree.

>
> >The root cause of the problem is the Debian/Ubuntu ecosystem is moving
> >away from the "universal OS" mantra and toward being a GNOME
> >bootloader for tablets and everything else can just go away. Thats
> >the war... abandonment of entire industry sectors or weird init
> >decisions are merely a small battle, the war is the move toward
> >turning the OS into a GNOME bootloader for tablets.

And we'd like to take over the "universal OS" mantle. When we have the
resources.

-- hendrik
k