Pretty much exactly 19 years ago I got on a train to Oxford and made Mark Shuttleworth's laptop successfully suspend and resume using ACPI and that was the turning point in my entire career
It was August or September 2004 and Ubuntu had been announced earlier that year at Debconf in Brazil but nothing had publicly shipped yet. I got invited to the internal dev summit that was happening in Oxford a couple of months after the announcement. At this point, nobody outside Canonical had really seen what was being developed.
@mjg59 It was 2004? Don't know why I thought it was earlier, as a Debian user still remember the "Ubuntu it's an ancient African word that means Can't install Debian" It's almost 20 years but I thought it was before the 2000's
Earlier that year I'd bought a Thinkpad X40 and through sheer bloody mindedness had figured out how to get it to successfully suspend and resume using the new hotness (ACPI) rather than the old and busted APM. APM was, with hindsight, an awful idea, and the Linux community stuck to it for far too long. That's a separate conversation.
I hung out in Oxford for a week and we spent the evenings watching awful things like Antitrust (thanks @Migueldeicaza). But I also managed to bodge many people's laptops into successfully suspending and resuming via ACPI, even if I'm not proud of the solutions invoked at the time
@mjg59@Migueldeicaza Is bodge your technical term? I think your techno-elite friends are losing their minds in too many rabbit holes. Here's how I resolve suspend resume:
(Worst case example: the VESA BIOS extensions gave you a mechanism to save graphics hardware state, and by marshalling that through userland we could save it before suspend and restore it after resume and then switch to the X server and all of this is awful but it made a bunch of machines work before we had graphics drivers that could do this for us)
There was then the digression of the naked people wallpapers, which everyone seemed to agree was a bad idea but nobody was really willing to actually fight Mark about, although I think @jeffwaugh made a far stronger representation than anyone else and deserves recognition for that
Anyway the Ubuntu 4.10 beta shipped and more laptops suspended and resumed than any other Linux distribution up until that point and Mark tried to hire me and I said that I was going to go and finish my PhD in fruitfly genetics and that was probably actually the right call?
@mjg59 I am on my last couple weeks; but I remember when I started 17½ years ago, you came by to swap laptops with elmo while he was doing my induction. It was a time when we still tried to get disk hibernation working...
But we ended up spending a lot of time just getting laptops into the hands of community members and fixing the bugs that they were hitting and I genuinely believe that this ended up changing the way people thought of Linux on laptops - it should work, and if it doesn't then that's up to the distro not the laptop vendor
I gave a talk on this sort of thing at OSCON in 2007 and a bunch of people in the audience were wearing t-shirts saying "I won't fix your laptop, but Matthew Garrett will" and I have no idea whether any of these still exist
Anyway I just wanted to fix my laptop and that's one of the reasons Ubuntu ended up dominating desktop Linux and a lot of history is entirely accidental The End
(Epilogue: I ended up resigning from Ubuntu for reasons that were a mixture of just being in a bad place at that time and a specific set of circumstances, and accidentally ended up Cc:ing a public mailing list, and also did so from my jailbroken iPhone while drinking beer at FOSDEM, so welp probably the first person to publicly ragequit Ubuntu with a proprietary user agent string)
@mjg59 is that resignation still public? In light of how canonical changed over the years it'd be interesting to compare whether the problems aren't in a way still the same.
@mjg59 we were rolling our own Debian-based distribution for our computer refurbishing project from about 2001 to 2006. We switched to Ubuntu in 2006 because it was lightweight enough to run on the machines were building, and, most importantly, it made installing proprietary drivers easy (less work). When Canonical decided to switch to Unity, we switched to Xubuntu. (Unity wouldn't run on a lot of the laptops we were refurbishing in 2010). Still using Xubuntu today at our project.
@chaslinux
> When Canonical decided to switch to Unity, we switched to Xubuntu. (Unity wouldn't run on a lot of the laptops we were refurbishing in 2010). Still using Xubuntu today at our project
Have you had a play with @trisquel? It's a deblobbed Ubuntu with a very lightweight but still human-friendly Mate desktop. It became my go-to when the Amazon Lens debacle make it untenable for me to keep using or endorsing Ubuntu.
@ben@mjg59 At some point, it became a supported path inside IBM to use a laptop (Lenovo) with Linux as your company computer. It was RHEL though. But only certain Lenovo models were supported. I wonder if that is related to those Linux friendly models.
@acsawdey@ben@mjg59 I had official blessing from IBM IT at one point to build a distribution on top of Ubuntu for use inside IBM. It was self-supported but officially accepted. No idea if that still exists now, I think they were pushed RHEL and fedora more
@sldrant@acsawdey@mjg59 Likewise I was part of the (small) group running Fedora. But we were mainly in the Development Labs, the service folk all HAD to run RHEL
For my part the end of the road came when IBM would let you select a MBP, I jumped ship for my work laptop. I do have a home server running Ubuntu as well as a one at an animal rescue (ZoneMinder and a bunch of IP cameras).
@mjg59 Do you think "it should work, and if it doesn't then that's up to the distro not the laptop vendor" still holds true?
Even though Linux is now much more widely adopted on personal computers, oftentimes ODMs don't care that much and only focus on firmwares and drivers for Windows (or with undocumented tricks only made to work a bit better on Windows). When we spot problems (such as ACPI errors actually), they always shrug and claim that it works on Windows and that's the end of the story…
@pieq my position has always been that if Linux doesn't do ACPI in the same way Windows does, it's a Linux issue. There's no realistic way for vendors to support two operating systems that behave differently.
@pieq many of the cases of divergence weren't explicitly addressed by the spec (eg, behaviour of concurrent access under certain circumstances caused deadlocks on some systems) so yeah figuring out what Windows did was the best plan
@mjg59 I'll bet a lot of people at that meeting had new laptops needing fixes because everyone's laptop (except for mine) was stolen at Ubuntu's previous Oxford meeting in August
(I heard from Mark on the icebreaker in February that same year. Also never worked for ubuntu but I was on the no-name-yet list so.)
@Ailuridae conceptually ACPI is about giving the OS the information and mechanisms it needs to configure and power manage the hardware, which is a solid plan. Whether any given implementation is actually good is a separate question.
@mjg59 Somewhat alarming to think that 19 years on, ACPI is still a mysterious spooky haunted forest festooned with bear traps, tar pits and other traps for the unwary... <watches flood of ACPI errors scrolling by as 6.5.x kernel boots on 2023 model laptop from Well Known Tech Company>
Add comment