On 6/10/2020 2:59 PM, Yves-Alexis Perez wrote:
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA256
<snip>
My plan before this conversation came up was to keep an eye on
what fixes were needed to get things working on the
Ubuntu/Fedora/RHEL front and then once those were upstream work on
getting those pulled into Debian.
I really think most of the work should be done in upstream projects
(so with upstream contributors), then making sure those changes are
included in the relevant Debian version. That also means really early
hardware shouldn't be needed since most of the *porting* work should
have been done (and actually mostly by Intel and maybe AMD engineers,
these days).
Agreed
I'd say this is mostly about having the correct fixes in the Linux
kernel, but maybe there are other parts which need fixes (maybe Xorg,
pulseaudio, stuff like that?).
I used to work on network switches and moving to PCs has been
eye-opening. It's amazing the amount there is that can break :)
- Audio is a constant headache (instead of earache).
- Graphics is frequently fun. Nvidia cards definitely add a challenge
but I think that's improved a lot. OLED panels are throwing up some issues.
- Power management - suspend and resume, energy certification. Hibernate
right now seems broken/not supported well which I suspect is going to be
an irritant to a lot of users.
But ethernet, wifi, bluetooth, touchpad, thunderbolt, USB, touchscreen,
card readers, thermal, secure boot, apps, firmware, fingerprint,
camera...the amount of stuff packed into these devices still amazes me -
and there is always something that crops up.
If you have experience on what was really needed in recent ThinkPads
it might be useful to reach out to relevant teams. I only have
hands-on experience with X250 (documented in the link I gave on my
first mail), which is a bit old now I guess (I have hands-on
experience on an X280 but it was maybe 18months after it was released
so basically everything was already working just fine)
We're going though this on our 2020 platforms right now so it's an
interesting point. I think I can safely say that it is constantly varied
:) The impression I get though is this year things are going more
smoothly than last year - I'd like to claim that this is because we are
more experienced.....I might be kidding myself.
I should document it though - right now my gut feeling is I can't point
to one or two candidates and go "they always cause trouble". It's truly
varied
<snip>
Anyway - if there was interest we could explore what was involved
with choosing a couple of platforms, getting them to a Debian
developer or two and going from there. I don't know how early in
the process we'd be able to make HW available - it is *really* hard
to get hold of these early systems (based on personal experience).
That's a challenge I'm willing to take on if it's something that
there is interest in.
So right now what would be the target hardware / timescale we'd be
talking about? There's definitely no reason that Lenovo product(s)
roadmap(s) and Debian stable roadmap should be aligned, so there are
few interesting things:
An interesting question. I'd actually like to bounce that back at the
community. After all - if the aim is to have platforms that work with
Debian the people who are going to want these are likely to be Debian
folk (you're not the first choice for Linux noobs ;))
I have some ideas but is there a 'wishlist' or guidance on which
platforms are the most popular? Either Lenovo specifics (makes my life
easier) or general "it should have at least xxxx".
Anybody want to setup a Debian survey? :)
- - making sure “current” generations products work fine on Debian
stable/Buster, so they could be “qualified” to ship with Buster
preinstalled (I'm unsure how realistic it for current products, but
making sure *they* work is a first start I guess) - - making sure
“next” generation products work fine on Debian stable/Buster so they
could be qualified to ship with Buster preinstalled (maybe a little
more realistic but I guess it also depends on the timescale) - -
making sure “future” products work fine in Debian testing (Bullseye)
so once it's released the products can be qualified on it and
hopefully ship with Bullseye preinstalled
This makes sense but I'll be honest - I haven't run Buster on anything
in a long time. For me too much doesn't work (touchpad, graphics, audio,
networking usually). I go straight for Bullseye and then to unstable and
sometimes to experimental.
For “current” products I guess we mean everything until the 2020
generation (so including X13/T14/T15, I don't really know that much
the rest of Lenovo lineup). For those I assume there would be no NDA
needed (and actually there might already be some hardware in the
hands of relevant people). For “next” I guess it depends a bit on the
timescale and the level of upstream support already here. For
“future” it's even more true, and my guess would be that most of the
stuff need to be done by Intel (and AMD) people first.
Yep.
So we have a few platforms coming out in the next few months. X1 Carbon
is usually high on users wish list - but maybe that's less the case for
developers? (I personally quite like mine...)
The P series with the mobile workstations are nice machines too - but
they can be beasts.
So I'd say “current” and maybe “next” are the most realistic if you
want to start somewhere.
I'm wondering if I/we can choose a couple of "next" and try that and see
what happens - but I need to go and shake some branches internally and
probably have some conversations off mailing list (I like doing
everything in public but...not sure Lenovo is quite ready to have all
plans in the open - maybe I'm overthinking it)
Full honesty disclosure - this month is nuts. I want to give some time
to this but please be patient if responses are slow. I wrote most of
this email yesterday but didn't get to hit send until....now
Mark
