The author listed quite a few apps then:

> A couple apps you won’t see on this list are Chrome and Slack

He explained we should:

> in my opinion you should boycott them until the developers learn how to write them to play nicer with memory

I stopped reading. Selection bias is selection bias, no matter how you try to talk it out.

This is either clickbait or outright absurdity apologist nonsense.

I currently have 19GB of swapfiles sitting in /var/vm/ on a maxed out 2015 MBP.

The new MacBook Pro is not professional and it is underpowered. Accept it already.

Would you mind indulging the audience (or me specifically) and outlining what you're doing that uses 30Gb+ of RAM? Is this one application with huge data sets, or many applications that must be run at the same time? Genuinely interested...

Not OP but this is mine right now and I'm not even doing anything that intensive. http://i.imgur.com/xYlR8Nw.png

Programmers stopped caring about memory not too long ago and this is the result.

Wow. I thought Eclipse was memory hungry, and given the comments here I also thought Sublime Text was frugal with memory compared with Atom.

For comparison, my development machine with a Docker environment and dev tools running: http://imgur.com/a/sic5M

I keep seeing Google Chrome Helper over and over and over in these screenshots. And sure, none of them are the biggest single memory user, but...

In my experience, quitting Chrome lowers the memory usage of kernel_task dramatically, so that Chrome is taking up even more memory that you'd think adding up all of the many Chrome-related processes.

I fully believe the OP, simply because they're not running Chrome. And I fully believe each of the posters above me, too, and I'm pretty sure they're all running Chrome.

I guess I'm suggesting that the difference is Chrome.

This was mentioned multiple times: https://news.ycombinator.com/item?id=12843261

Try running a few docker containers to support development work. A few DBs, messaging servers, Lucene indices, etc. Try prototyping applications using map-reduce for huge datasets.

"Pro" doesn't always refer to the artistic line of professionals. Many developers see this as a huge problem, and there are a lot of us.

Our options now include the 2005-era looking systems at system76.com, running Linux. I love Linux for servers, but I really don't like the GUIs, or acting as my own systems integrator. I want a true modern "Pro" macOS system, and that doesn't exist right now.

> Our options now include the 2005-era looking systems at system76.com, running Linux.

Or in fairness a Dell XPS15 which aside from supporting 32GB of RAM (and is orderable with that) has the quad-core i7HQ vs the Macbook Pro's and it's about 500 quid cheaper in the UK.

>Try prototyping applications using map-reduce for huge datasets.

Yep, a Spark job I was running recently used up my Mac's 16GB rather quickly.

> Try running a few docker containers to support development work.

Docker containers should not take more memory than the same app running without docker + a few megs for duplicated libraries. What matters is what's running in the containers.

Doesn't Docker on Mac run the containers inside a Virtualbox VM running Linux? That should take quite a bit more memory than just running the app directly.

It used to, but the newer Docker for Mac uses xhyve which is more lightweight. It also by default runs a pretty stripped down version of Linux. The default VM size is 2Gb, but you can increase that if you need to.

I'm currently working on a project which involves a Docker replication of a production environment. The Docker Compose environment has 7 docker containers including Cassandra, Kafka and Flink plus some nodejs micro-services. This all runs happily in a 6Gb docker container leaving memory free for Eclipse, Atom, Outlook and a bunch of other applications on a 16Gb 2015 MacBook Pro. At no point have I felt that the machine is performing poorly due to memory exhaustion.

I'm comparing running on linux+docker vs running on linux. Virtualisation is present in both. If you've got the possibility to run the same app directly on OSX and you're not preparing for deployment on another system... why wouldn't you do that instead?

Docker for Mac uses the built-in Hypervisor.framework through xhyve. Still virtualizating Linux, though, so there's some overhead.

> Try running a few docker containers to support development work

If I don't do that, then am I not a professional?

Let's just ignore very popular apps I don't like that happen to use a lot of memory so I can make a point.

And Firefox with one single website. But even so, the test is interesting because others can run a more realistic use case and report on that.

Maybe they shouldn't use tons of memory carelessly.

But they do. And not necessarily carelessly.

There seems to be a pretty consistent pattern of Chrome and apps built with Electron having large memory footprints. You should turn up here when an Atom article hits the front page and see how HN commenters' attitudes to memory consumption change then...

Not saying we have to, and the memory constraint might eventually make it infeasible, but my team's developer process is a lot easier when we can run our stack (12 web services, db, message bus, reporting service) at once.

There's definitely some optimization opportunities to address, but since the only immediate bottleneck that's causing is in local dev, our resources are better spent buying beefier dev rigs and focusing on delivering value.

My machines are currently 16GB each (I have two MBPs). I imagine increasing the amount of RAM I use in my next machine upgrade.

Anyone working on large scale data problems or doing advanced analytics (Spark, R, Python, etc.) would appreciate having more memory to allocate to their problems.

The analysis in this blog underestimates the complexity of various compute environments.

Mine: 14.6GB used, 18GB swap used.

Twitter 2.99GB (old, ad-free client) Java 1.22GB Photoshop 1GB Transmit, Spotify, Sublime, etc - 700MB each. Chrome core 2.8GB and then 79x (!) Google Chrome Helper, each anything up to 2GB. Obviously not listing the loads of things in the 200-700MB range.

Currently have Firefox, Safari and VirtualBox closed, though they are often open for testing.

similar point here: https://blog.rinatussenov.com/leave-apple-alone-538d7619ce9e...

I tend to use AWS Windows or Linux machines for work and use the Mac basically for Web browsing, RDP client and occasional Microsoft Office. This has removed my need for a desktop or notebook with more 16MB, effectively lenghting the useful life of my hardware. If I feel I need more RAM (up to 2TB), I just shutdown the AWS instance, change the instance type and start it again. Actually, with this setup, I very rarely have the urge to use my laptop, I have a desktop in the 3 places I usually work from so I'm freed to drag a laptop with me unless I travel.

I used a chromebook for about a year, and would RDP to my home desktop when I needed anything beefier... after that I went back to a new rMBP because I needed macOS at the time. The 16gb is rarely an issue for me, but I could easily see certain classes of workload that could make it so.

None of the memory consumptions are realistic, I use most of the listed apps. E.g. Xcode using 300 MB? Do you have one file open?

Disk cache buffers. It's what's left after all your applications are resident, and often impacts IO performance.