I've been in the market for a new laptop for a long, long time. I bought my last laptop in 2011 and have run Linux on it, but I've been very disappointed with the hardware (a ThinkPad X220T). I had dreams of using the Wacom tablet built into the screen to make artwork, but it just didn't work out, and it wasn't cheap either. It was time for a change. Maybe every other laptop can be for Linux. Sorry, Linux, it's not you, it was the hardware.

Ideally I want a setup at home similar to what I have at work: a MacBook Pro with a large monitor and an external keyboard. Ergonomics are important to me and at least for the Java I hack on at work, being able to see more code at once helps a lot. However, at home I'm quite space constrained so the only thing from above that I got was an external keyboard. Instead of an external monitor I bought a Curve Flex stand to get the screen much closer to eye level.

As for the laptop itself I went for an M2 Macbook Air. I was especially focused on increading the display resolution from 1366 by 768 pixels on my old ThinkPad. I did consider a Linux-based Dell XPS but the resolution was too low. The base prices are below but I did double the RAM to 16 GB.

Name          	display resolution	diagln	weight	price
old Thinkpad	1366 by 768 pixels
MacBook Air M1	2560 by 1600 pixels	13.3"	2.8 lbs	$ 999
MacBook Air M2	2560 by 1664 pixels	13.6"	2.7 lbs	$1199
MB Pro 14" M2	3024 by 1964 pixels	14.2"	3.5 lbs	$1999
MB Pro 16" M2	3456 by 2234 pixels	16.2"	4.8 lbs	$2499
Dell XPS 13    	1920 by 1200 pixels	13.4"	2.6 lbs	$ 989

This is my first experience with an ARM processor in a computer and it's… interesting. The first Apple silicon chip in a laptop was the M1, of course. A colleague at work has one and he hasn't had much trouble that I know of.

Most apps are installing just fine on the M2. Firefox, Element, VS Code, and Inkscape don't seem to have any trouble. (Inkscape does have a warning about Ventura, macOS 13.0, not being supported, but I haven't hit any problems yet.) I did open a bug for Quiet not working on the M2 without Rosetta.

Containers are keeping me busy. On my Mac at work I've been using Docker for years but I thought I'd give Podman a try. Between Podman, the M2, and my grasp of only basic Docker commands, I've been struggling a bit. The fact that I'm trying zsh instead of bash is also slowing me down here and there. I did alias docker to podman, as instructed.

docker run hello works just fine and shows me some cute seals:

!... Hello Podman World ...!

         .--"--.           
       / -     - \         
      / (O)   (O) \        
   ~~~| -=(,Y,)=- |         
    .---. /`  \   |~~      
 ~/  o  o \~~~~.----. ~~   
  | =(X)= |~  / (O (O) \   
   ~~~~~~~  ~| =(Y_)=-  |   
  ~~~~    ~~~|   U      |~~ 

Great! Then I moved on to docker-aio (all in one) from Dataverse and had problems that I'll probably worry about on work time.

A much smaller Docker is example is from my notice project where I use a Tidyverse image from Rocket rather than installing all that R stuff on my laptop. Here's the warning/error was getting:

% docker create --name extract pdurbin/notice WARNING: image platform ({amd64 linux [] }) does not match the expected platform ({arm64 linux [] })

After much poking around I made a commit that seems to help. It amounts to adding --platform linux/x86_64 here and there, which I picked up from a Rocker issue. I did smile at a comment in another issue:

Isn't Docker supposed to prevent just this kind of problem?

Yes, you'd think so.

Now that I'm writing up this blog post and hacking some more I'm realizing that I may not need that --platform flag after all. The last step of copying out the image was failing this this:

% docker cp extract:/results/figure.png /tmp
Error: 2 errors occurred:
* copying to host: copier: put: error resolving "/tmp": open /tmp: too many levels of symbolic links * copying from container: io: read/write on closed pipe

The fix was pretty simple. In the docker cp I changed /tmp to /tmp/out (it could be any subdirectory of course). I'm guessing this is necessary because I switched to Podman. Like I said, lots of change. New chip architecture, new container software, new macOS version. I'm not surprised I'm hitting some bumps.