Re: Anyone using stretch/buster/sid on ARMv4t ?
- Date: Tue, 7 Nov 2017 12:00:42 +0100
- From: Adam Borowski <kilobyte@xxxxxxxxxx>
- Subject: Re: Anyone using stretch/buster/sid on ARMv4t ?
On Tue, Nov 07, 2017 at 11:16:41AM +0100, John Paul Adrian Glaubitz wrote:
> I think a possible solution is the plan we had inside Debian Ports which is
> to introduce a Britney instance within Debian Ports and hence be able to
> provide a Debian testing release.
> My dream would be to not to have the distinction between release architectures
> and ports architectures, but rather something like Tier I and Tier II
> architectures with the Tier II architectures sharing the characteristics of
> the Tier I architectures but without any support and without the buildds
> and porterboxes being maintained by DSA.
It would be great -- I tried to make an unofficial Jessie release for x32,
but doing the equivalent of Britney turned out to be too hard. The main
reason was binNMUs: any out-of-archive binNMU conflicts with official
binNMUs that come later, there's no record of in-archive binNMUs that's
reasonably accessible to an outside observer (at least a non-DD, I wasn't
one at the time).
The main reason I can't recommend x32 to any serious user, even for its
primary role of running inside containers -- no even unofficial security
support makes it a non-starter. On the other hand, as only very few
packages have ports-specific changes (ie, a +x32 version in unreleased),
rudimentary security support would be mostly a matter of automated builds
of whatever hits the security repo. Embargoed information is not supposed
to reach non-DSA machines but a slight delay wouldn't be unacceptable.
It's a chicken-and-egg problem: no testing release means no good way to
snapshot it as unofficial stable, no stable means no users, no users means
no motivation to work on the port.
. I for one consider using x32 with a GUI to be a waste of time: the
gains are significantly smaller memory use but only modest speed gains. All
modern x86 have so much memory that saving some is not worth extra human
effort for just a single instance, thus reasons to use x32 are limited to
running many containers on one machine (where memory savings scale) and
possibly serious number crunching (where even a small speed gain matters).
⢀⣴⠾⠻⢶⣦⠀ Laws we want back: Poland, Dz.U. 1921 nr.30 poz.177 (also Dz.U.
⣾⠁⢰⠒⠀⣿⡁ 1920 nr.11 poz.61): Art.2: An official, guilty of accepting a gift
⢿⡄⠘⠷⠚⠋⠀ or another material benefit, or a promise thereof, [in matters
⠈⠳⣄⠀⠀⠀⠀ relevant to duties], shall be punished by death by shooting.