This proposal is for use with #43651. We propose using type parameters to add compile-time-type-safe variants of sync.Pool, sync.Map, and atomic.Value. This proposal is not targeted for any specifi...

Happy to see this implemented (even as a WIP/Draft). Also take a note how small patch is. @Tailscale rocks as always but don't bring this to your prod, for now 😄#golang twitter.com/commaok/status…

I hacked up a simple way to assert that a Go value is immutable. It's pretty raw and certainly has bugs, but it appears to work: github.com/tailscale/go/p… (And big hat tip and thank you to @Tailscale, where amazingly enough this was part of my day job today.)

Should work with normal -race if reflect.Freeze does fake read I think. This would allow to use existing -race CI config rather than a new build. twitter.com/bradfitz/statu…

Proposal: Soft memory limit Author: Michael Knyszek Summary I propose a new option for tuning the behavior of the Go garbage collector by setting a soft memory limit on the total amount of memory t...

Sometimes to get good speedup you just need to read the code, thanks @ignaciohagopian #golang twitter.com/ignaciohagopia…

Two low-hanging fruit optimizations in strings.Replace that allows double-digit speedup and memory saving in the existing reference benchmark. #golang #performance go-review.googlesource.com/c/go/+/343089

This past summer, @markgritter spend 25 days battling the Go garbage collector—and emerged victorious. Good news: WE DID NOT HAVE TO REWRITE OUR CLIENT IN RUST. 🙅🏻‍♀️🦀 Read this post for Mark's epic journey to reducing the multi-gigabyte memory spikes. a…

Here’s how we went from using 1400KB of RAM per repo to just 310KB without affecting latency.

As discussed in #33283, Go's implementation or valueCtx.Value is recursive which means that the cost grows linearly with the depth of the context chain. But one issue not discussed there is tha...

A story in 3 acts

Tailscale on iOS runs as a special kind of app, a Network Extension. This lets us run in the background, so we can secure traffic from all of your applications, without them having to change anything. But with this power comes a memory straightjacket. Normal…

This is an umbrella issue for supporting feedback-guided optimization. This has been discussed offhand in a few other contexts (inlining decisions: #17566, code layout decisions: #20356, language c...

Please take the 2021 Go Developer Survey. We want to hear from you!

@felixge I know VS Code enables you to run > "Go: Toggle gc details" to see some more details, but nothing like this - though it would be cool!

cmd/benchstat: new version of benchstat Remember that if you're doing benchmarking, use benchstat instead of benchcmp. #golang go-review.googlesource.com/c/perf/+/30996…

Here's a list of some changes to look forward to in #golang 1.18! The first beta should be out in a month, and the final release in three months. Each change with a link, so you can read more.

The compiler currently compiles a+b+c+d as a+(b+(c+d)). It should use (a+b)+(c+d) instead, because the latter can be executed out of order. More broadly, we should balance trees of associative comp...

I run this bot to collect benchmarks; anyone who's watching Go-1.18 release, we (MK) shrank the starting heap size, which is superficially good, but lots of small benchmarks ran entirely within the old initial heap and now they don't -- our metrics went kerflooie.…