> He needs to stop fighting with the winds he can't control. Users gonna be users, and people gonna be people. Everyone won't be happy, never ever.
Right - but it kinda sounds like he's facing headwinds in a lot of different directions.
Headwinds from Apple, who are indifferent to the project, stingy with documentation, and not inclined to reduce their own rate of change.
Headwinds from users, because of the stripped down experience.
Headwinds from the kernel team, who are in the unenviable situation of having to accept and maintain code they can't test for hardware they don't own; and who apparently have some sort of schism over rust support?
Be a heck of a lot easier if at least one of them was on your side.
> Headwinds from Apple, who are indifferent to the project, stingy with documentation, and not inclined to reduce their own rate of change.
That is part of the challenge he chose to take on.
> Headwinds from users, because of the stripped down experience.
Users can be ignored. How much you get users to you is your own choice.
> Headwinds from the kernel team, who are in the unenviable situation of having to accept and maintain code they can't test for hardware they don't own
You don't have to upstream. Again, it's not the kernel team that chose to add support for "hostile" hardware so don't try to make this their problem.
> and who apparently have some sort of schism over rust support?
Resistance when trying to push an entirely different language into an established project is entirely expected. The maintainers in question did not ask for people to add Rust to the kernel. They have no obligation to be welcoming to it.
> Be a heck of a lot easier if at least one of them was on your side.
Except for the users all the conflicts are the direct result from the choice of work. And the users are something you have to choose to listen to as well.
"Their boss" - I'm not sure that boss is best word here.
"did ask for it" - did he? Because from my perspective it looks more like he gave the bone for corporations so they will shut up for rust in kernel. After some time it will end up "Sorry but rust did not have enough support - maintainers left and there were issues with language - well back to C"
I don’t think that’s an accurate way to describe what happened, no. He seems to be enthusiastic about it and to genuinely want it to succeed.
> "A lot of people actually think we're somewhat too risk averse," said Torvalds. "So when it comes to Rust, it's been discussed for multiple years by now. It's getting to the point where real soon now, we will actually have it merged in the kernel. Maybe next release."…
> "Before the Rust people get all excited," the Linux kernel creator and chief said. "Right? You know who you are. To me, it's a trial run, right? We want to have [Rust's] memory safety. So there are real technical reasons why Rust is a good idea in the kernel…”
> “And hopefully, it works out, and people have been working on it a lot, so I really hope it works out…”
Last September he was still insisting he thinks the project will not fail, and he was not exactly subtle in his criticism of maintainers who refuse to engage with it in good faith.
> "Clearly, there are people who just don't like the notion of Rust, and having Rust encroach on their area.
> "People have even been talking about the Rust integration being a failure … We've been doing this for a couple of years now so it's way too early to even say that, but I also think that even if it were to become a failure – and I don't think it will – that's how you learn," he said.
> "So I see the whole Rust thing as positive, even if the arguments are not necessarily always [so]."…
> With impressive diplomacy, considering his outbursts of years past, Torvalds went on, "There's a lot of people who are used to the C model, and they don't necessarily like the differences... and that's ok.
But yeah, I still don't think it's all that inaccurate: He may not have wanted it to fail, and still not think it's a technical failure... But socially? Still seems possible he'd be starting to think that while the Rust language per se is a technical success, all the drama surrounding the integration of it into Linux means that that is turning out to be a social failure.
(Or maybe I'm just projecting because that is what it looks like to me.)
Another uphill battle that I haven't seen anyone mention is just how good mobile AMD chips got a year or so after the M1 release. I wouldn't buy a Mac to run Linux on it when I can buy a Lenovo with equally soldered parts that'll work well with the OS I wanna run already.
A lot of it is simply AMD getting on newer TSMC nodes. Most of the Apple's efficiency head start is better process (they got exclusive access to 5nm at first).
That's my understanding as well, as soon as the node exclusivity dropped they were ballpark equal.
Many ARM SOC are designed to run on battery only so the wireless packages and low power states are better, my AMD couldn't go below 400mhz.
But yeah the "Apple M hardware is miles and leagues away" hypetrain was just a hypetrain. Impressive and genuinely great but not revolutionary, at best incremental.
I hope to be able to run ARM on an unlocked laptop soon. I run a Chromebook as extra laptop with a MediaTek 520 chip and it's got 2 days battery life, AMD isn't quite there yet.
> But yeah the "Apple M hardware is miles and leagues away" hypetrain was just a hypetrain. Impressive and genuinely great but not revolutionary, at best incremental.
It's more nuanced than that. Apple effectively pulled a "Sony A7-III" move. Released something one generation ahead before everybody else, and disrupted everyone.
Sony called "A7-III" entry level mirrorless, but it had much more features even when compared to the higher-end SLRs of the era, and effectively pulled every other camera on the market one level down.
I don't think even they thought they'd keep that gap forever. I personally didn't think it either, but when it was released, it was leaps and bounds ahead, and forced other manufacturers to do the same to stay relevant.
They pulled everyone upwards, and now they continue their move. If not this, they also showed that computers can be miniaturized much more. Intel N100 and RaspberryPi/OrangePi 5 provides so much performance for daily tasks, so unimaginable things at that size are considered normal now.
I like the Sony story, but I don't think Apple did "pull everyone along" like that, they had an exclusivity deal with TSMC to be first on a good manufacturing node improvement. They took their high-quality, high-performance iPhone SoC, gave it more juice and a bit better thermals.
It's just another "Apple integrating well" story.
Their SoC is huge compared to competitors because Apple doesn't have to make a profit selling a SoC, they profit selling a device + services so they can splurge on the SoC, splurging on the SoC plus being one node ahead is just "being good", the team implementing Rosetta are the real wizards doing "revolutionary cool shit" if anything
> they had an exclusivity deal with TSMC to be first on a good manufacturing node improvement.
...plus, they have a whole CPU/GPU design company as a department inside Apple.
Not dissimilar to Sony:
Sony Imaging (camera division) designed a new sensor with the new capabilities of Sony Semiconductor (fab), and used their exclusivity to launch a new camera built on top of that new sensor. Plus, we shall not forget that Sony is an audiovisual integration powerhouse. They one of the very few companies which can design their DSPs, accompanying algorithms, software on top of it, and integrate to a single product they manufacture themselves. They're on par with Apple's integration chops, if not better (Sony can also horizontally integrate from Venice II to Bravia or Mics to Hi-Fi systems, incl. everything in between).
The gap also didn't survive in Sony's case (and that's good). Nikon and Fuji uses Sony's sensor fabs to use their capabilities and co-design sensors with the fab side.
Canon had to launch R series, upscale their sensor manufacturing chops. Just because Sony "integrated well" when looked from your perspective.
Sony is also not selling you the sensor. It's selling you the integrated package. From sensor to color accuracy to connectivity to reliability and service. A7-III has an integrated WiFi and FTP client to transfer photos. A9 adds an Ethernet jack for faster transfers. Again, integration within and between ecosystems.
>But yeah the "Apple M hardware is miles and leagues away" hypetrain was just a hypetrain. Impressive and genuinely great but not revolutionary, at best incremental.
Compared to the incremental changes we've seen the previous 10 years before it arrived on AMD/Intel space, it was revolutionary.
Was Intel switching from the Pentium 4 to the Core architecture considered revolutionary at the time? Was AMD's bulldozer architecture? I don't recall.
We must have different definitions of the word "revolutionary". They put a high-end mobile chip in a laptop and it came out good, what's revolutionary? The UMA architecture has advantages but hardly revolutionary.
The jump in performance, efficiency, battery time was not incremental or "evolutionary". Such jumps we call evolutionary.
What they did doesn't matter. Even if they merely took an intel laptop chip and stuck a chewing gum on it, the result was evolutionary.
So much so, that it put a fire under Intel's ass, and mobilized the whole industry to compete. For years after it came out the goal was to copy it and beat it.
What did you expect to call "revolutionary"? Some novel architecture that uses ternary logic? Quantum chips?
And some of these Lenovos are relatively upgradable too. I'm using a ThinkPad I bought refurbished (with a 2 year warranty) and upgraded myself to 40 GB of RAM and 1TB of SSD (there's another slot too if I need it). It cost me $350 including the part upgrades.
Prices seem to have risen a bit since I bought mine. Here's a similar model with a Ryzen 5 7530U for $355: https://www.ebay.com/itm/156626070024 It is certified refurbished and has a two year warranty. It has a SODIMM slot and supports dual SSDs, although not the full size M.2.
Right - but it kinda sounds like he's facing headwinds in a lot of different directions.
Headwinds from Apple, who are indifferent to the project, stingy with documentation, and not inclined to reduce their own rate of change.
Headwinds from users, because of the stripped down experience.
Headwinds from the kernel team, who are in the unenviable situation of having to accept and maintain code they can't test for hardware they don't own; and who apparently have some sort of schism over rust support?
Be a heck of a lot easier if at least one of them was on your side.