I live a double life: Apple for laptops and Linux for desktops (and I'm mostly a Desktop user who manages a lot of Linux servers).

Both platforms have their pluses and minuses. I like to carry ideas between them for my development and environments. Found this extremely fruitful for me over the years.

We ordered a bunch of M1 airs for portability and power efficiency. Will see how it goes.

I personally love both Linux and Apple for laptops but, for the out of the box experience and reliability, Apple is more polished due to integration. My EliteBook has no problems with Linux so far but, when a flawlessly running driver degrades after an update (cough, intel e1000e, cough), it leaves a bad taste in the mouth. OTOH, I'm firstly a Linux guy who prefers and writes GPL software.

Of course YMMV.

Yep. Mac's look like a comfortable unixy environment. Preinstalled zsh, a fast terminal emulator (item), gnu utils. I'm sure I could make it work as a dev machine.

I have to admit that m1 macbook air battery life is compelling. A long time ago I was thinking about a perfect laptop that would be very light and have a 20+ hours of battery life. This comes close. Even though I have never used a laptop more than 5 hours on battery. And that happens a couple of times a year.

Hope you'll enjoy your m1 air.

Thank you. I’m really excited about it. Using a different CPU architecture is always exciting. :)

Using Mac is like driving an Audi S8. Comfortable, smooth, powerful. However, it's not your hand tuned Impreza which can read your mind while going above 200Km/h.

Even an Intel MacBook pro can go a long time on battery (I have a personal Mid2014, configured all-out). I've developed some system-abusing scientific code on it and, it really delivers.

The trick for me is, I can make a Mac dev machine without modifying it with homebrew, et. al. I install a Linux VM to VMWare, give it a static IP, install all required tools, servers and services on it. On the macOS side, I use Eclipse for IDE and use clang/llvm for compiling (since I also aim to code which behaves same on both gcc and llvm).

I develop on macOS and interface it with Linux VM via network if it's absolutely necessary. Linux VM also hosts some heavy tooling like LaTeX which cannot be installed/updated on macOS very cleanly (I know macTeX. It doesn't play nice with newer macOS, due to Apple's locks on the OS).

Then everything becomes transparent for me. Pull -> Develop -> Push in either direction. Eclipse already syncs itself via oomph and cloud. Code is portable, environment is same. It also ensures code compatibility, allows me to see compiler effects and run tests on many platforms.

As aforementioned, it also inspires me to write better software. My code carries Apple's sensible defaults and it-works mentality with Linux's flexibility options. This approach allowed me to create a series of utilities for a project in limbo. These small no-setup utilities saved a lot of labor and ultimately saved the project. So yes, apple is a walled garden and it's not optimal but, they do a lot of right things. We can selectively carry them to more open platforms to make them better. Similarly, open platforms' flexibility can be carried to some macOS applications so, the environment better accommodates power users out of the box.

Homebrew and other tools are nice but, I don't like to shoehorn stuff which doesn't fit natively into an ecosystem.

> I know macTeX. It doesn't play nice with newer macOS, due to Apple's locks on the OS

I’ve been using LaTeX from /opt/local with zero integration issue, even on Big Sur. What are the issues you’re alluding to?

Otherwise, I agree with your take in general, except that I do everything you do in a Linux VM directly in macOS. I develop and run my codes on a Linux workstation and an iMac, and going back and forth improves the codes a lot. Performance and usability issues are spotted much earlier.

I much prefer Macports’ approach than Homebrew’s. It is nicely self-contained in /opt/local, does not break when a system library is updated, and the file system layout is much saner. Just update your environment in zshrc and you get all the benefits whilst not changing anything for all the stuff that you don’t run in a terminal.

In the past, MacTeX team had a problem with a particular OS release (when SIP was released and enabled) and, I was in the middle of my Ph.D. At that time, had no time to wait for problems to resolve. I got VMWare, installed Linux, tuned my LaTeX environment and never looked back.

I'm sure that all the problems are resolved by now but, my workflow is mature now and everything is working flawlessly. Considering I'm going to need Linux anyway, did no efforts to move my LaTeX workflow back to macOS again.

Since the code I'm developing is going to be used in a lot places, I'd rather develop it in two distinct environments and run tests on each. Also, I like to experiment with different development tools on different environments. Experimenting and experiencing each environment broadens my horizon. Also it's more enjoyable IMHO.

Didn't play with Macports TBH. I don't think I'm going to use it but, will take a look to it.

Another thing is, I don't customize/change my terminals much. When you manage 1000+ servers with a team, customizing each terminal to your liking is not feasible so, I can work pretty fast with stock bash or anything. I'm old school and don't like flashy console setups anyway. :D

> In the past, MacTeX team had a problem with a particular OS release (when SIP was released and enabled) and, I was in the middle of my Ph.D. At that time, had no time to wait for problems to resolve. I got VMWare, installed Linux, tuned my LaTeX environment and never looked back.

I feel for you; the middle of writing up a PhD is about the worst time to have this type of technical issues. I remember being afraid of any update back then.

> Didn't play with Macports TBH. I don't think I'm going to use it but, will take a look to it.

To me, Macports is the closest to a sane package manager like on FreeBSD or most linuxes. There is practically no learning curve if you’ve already used one. But yeah, it’s not flashy or cool like Homebrew and it wants to use the system software as rarely as possible so it will reinstall zlib etc. I think the trade off is acceptable because then everything is more stable and predictable as the exact libraries used are known and tested, and won’t be broken by an update.

> Another thing is, I don't customize/change my terminals much. When you manage 1000+ servers with a team, customizing each terminal to your liking is not feasible so, I can work pretty fast with stock bash or anything. I'm old school and don't like flashy console setups anyway. :D

Yeah. In terms of looks, have simple settings to see at a glance if I am on my local computer, a workstation over the LAN, or a cluster somewhere else (I like clean terminals). I have one git repo with settings files and zsh modules though. So I spent quite some time fine tuning everything, and all the computers I use behave the same, whether they run macOS or any Linux distro (or even Cygwin, actually).

> when a flawlessly running driver degrades after an update (cough, intel e1000e, cough), it leaves a bad taste in the mouth.

Hah! I see I'm not alone. For me it was the iwlwifi drivers. In one kernel revision, they worked fine; in another, they'd randomly panic and bring down the interface. Downgrading the firmware packages didn't do anything. Kernel upgrades have worked as of late.

Though, it may be hardware-related in my case. I've had nothing but trouble with that particular card. I was going to replace it, but it seems that Lenovo really likes Loctite on their M.2 screws and it won't budge. Tempted to use a soldering iron to melt the thread locker...

Intel went through a phase where they managed to break their graphics and network drivers at the same time.

I thought whether my hardware was failing but, when they broke three different NICs with the same update, I understood what happened.

Hope they don't do it again. I'm not going to use their processors in my next build but, I tend to like their NICs.

> I thought whether my hardware was failing but

You're now making me question whether this is a problem with the wifi card or the driver. I admit I'm still convinced it's probably the card (I have an earlier revision in a ThinkPad that works fine), but now I'm not so sure!

Curious: Are you talking about that "little" issue where the i915 module was randomly causing a panic? I ran into someone who had some really perplexing issues that a kernel update later resolved, and their dmesg was strongly hinting (to me) that it was linked to the i915 drivers. That, and a bug report I managed to dig up.

And I also agree. I'm looking at a Ryzen build, too. I have a couple of Intel NICs in my file server though and they're better quality IMO than the Realtek cards.

Actually it can be anything. My mother owns a HP Spectre X2 convertible with an Intel WiFi card. Only the driver bundled with the Windows works. Any newer version via Windows Update or Intel breaks the card after a standby-wake cycle. Both updated drivers support the card on paper but, either the driver has a problem or the way the hardware designed on that particular computer messes some stuff up. BIOS upgrade didn't fix anything too.

> Are you talking about that "little" issue where the i915 module was randomly causing a panic?

No, I started to see some lines in the display like a faulty GPU draws. Sometimes a line, sometimes small corruptions were visible but, it was always fixed itself after a minute or two. The laptop in question is extremely low power so, nothing gets hot.

Newer kernel updates fixed these issues too.

Intel's mid-range and higher-end NICs have proper offloading and processors so, they neither tax the system much or fail to reach their advertised speeds. They're real cards for real loads, so they're more reliable AFAICS. A Realtek card works reasonably well for most of the light loads but, load it up with sustained high load, it cannot saturate the interface as it should.

> Any newer version via Windows Update or Intel breaks the card after a standby-wake cycle.

hahaha that's insane.

I've seen some weird things come out of HP machines in the past, so somehow none of the above would surprise me.

> No, I started to see some lines in the display like a faulty GPU draws.

Ah, interesting. There was a drive bug in the i915 module that could potentially cause a panic on certain hardware. I never experienced it myself, but I ran into someone who had a similar issue. I don't think their problem was tied to the i915, but it was also resolved by a kernel update.

> A Realtek card works reasonably well for most of the light loads but

Yeah, and the Realtek chipsets are much more cheaply designed. Although, what puzzles me is that most of the Intel chipsets aren't that much more expensive (depends on vendor, of course, but you can easily find off-brand dual-port cards for ~$40USD with an Intel chip). It really makes it pointless to invest in Realtek cards that are both cheap and may or may not work at all.

I remember having an issue with the onboard one in my desktop where it would randomly disconnect after negotiating 1Gbps. If I forced it into 100Mbps it would be fine. But, as with the earlier discussion(s), that was also resolved with a kernel update. I used an Intel card for a few years because of that reason, so I have no idea when it was eventually fixed.

> hahaha that's insane.

Moreover, out of the box driver is a stock intel wifi driver. It's not vendor-specific. Phew.

> I remember having an issue with the onboard one in my desktop where it would randomly disconnect after negotiating 1Gbps.

That's how the new driver broke one of the NICs. Other ones were just failing to detect carrier at all. The bug you're referring is [0], which I also added some feedback.

[0]: https://bugzilla.kernel.org/show_bug.cgi?id=205047

Realtek's cards are not bad. They're not slow (in terms of latency). They're just not suited to sustained high loads (~60% of interface capacity).

> The bug you're referring is [0], which I also added some feedback.

Possibly not, unless I'm misunderstanding. The bug was seen in a Realtek card in 2011-ish.

I'm suspicious that might've colored my opinion of Realtek early on and why I still buy Intel cards to this day. Well, depending on the motherboard, of course.

Oh, then Intel actually caught-up with Realtek and broke their cards the same way for some time then. :)

Realtek's 8139's first revisions was not able to perform well in most cases (~30mbit max speed, unrealiable). 8169 is much much better. Their new wireless cards also work very well if the driver is good, but they have so many sub-models with very different feature sets, so you need to get the exact chip for your needs.

I use a 8111 (PCIe, Gigabit) at office for desktop to laptop bridge and it works fairly well, no stability or speed problems for now.

> Oh, then Intel actually caught-up with Realtek and broke their cards the same way for some time then. :)

Hadn't thought about it that way, but that's hilarious. Realtek's such a trend-setter.

> 8169 is much much better.

Just checked, and apparently that was the card in the machine I had issues with. I think part of the problem is that it was new at the time. So, I can't really fault the drivers per se. I never had any issues with it after they were fixed.

I have a Macbook and used to have a Hackintosh for my desktop. But this ARM move made me move away from that solution on the desktop.

I tried using Windows with WSL2 first but there was just too many workarounds I had to do for my workflow.

I then looked into the most popular Linux distributions and chose Pop!_OS. Suffice to say, I am really satisfied with it. It's probably the Linux distribution that I feel is the closest to MacOS that I have used.

I am interested in the M1 Macs but my 16" MBP is still fine. I will look into the ARM Macs next year to see what they offer.

The question is how far he can get without Apple's documentation. On the CPU side, probably pretty far as they use the standard ARM ABI. On the GPU side? This is where things will get really hairy. Same with the machine learning cores.

I'm not planning to get a macbook. My main contender would be an XPS with a new ryzen cpu on 5nm, if they'll have one next year, or a thinkpad again with Ryzen 5nm. I would not accept anything less.

But thinking about it, it might be that the mobile AMD CPUs are a year behind the desktop ones. So next year the best ryzen could still be at 7nm Zen3. In that case apple will have a 2 year transistor lead. But, will see.

Is there any statement Dell will build Ryzen XPSs?

Only wishful thinking from my side. As mentioned in the post, I hope that the M1 performance and aggressive apple marketing will push premium pc laptop manufactures in adopting the best performance x86 CPUs. And I also hope that AMD will seize the opportunity to get a good chunk of mobile market.

Of course, next year laptop models are probably being finished right now, and it could be that intel still found a way to bribe premium manufactures from including AMD in their premium models.

In my opinion they'll lose marketshare if that happens. But, there are most likely smart people with a lot of more information than me making these decisions.

Linux should be on it shortly, running in a VM environment.

That won't satisfy the purists, but it should satisfy the practical in most cases.