Post

I ran a test of container file system performance on Mac OS by the simple expedient of measuring `npm install` with a cold cache. github.com/llimllib/con... Orbstack did shockingly well, paying basically no performance tax whatsoever for running in a container. I've not previously used it

10/6/26, 8:36 pm0 reposts1 like

llimllib

@billmill.org

10 June

But I suppose I'll be giving it a try. Apple's container framework does somewhat better than colima, but you pay a pretty heavy performance tax to use either. This aligns with my experience - dev work with either of them is slow in node, python, or rust, where large dependency folders cause

llimllib

@billmill.org

10 June

a heavy performance tax when npm, pip, or cargo need to cross the filesystem boundary frequently.

llimllib

@billmill.org

10 June

It's far from a perfect test - the results depend heavily on the network, for example - but I think the results are consistent and well-separated enough to be convincing to myself at least. They also align with my prior biases, so they must be correct, right?