chore: prefer atomic.Pointer to atomic.Value

[arnehormann]

This improves the situation by closing down the api and enforcing the required type.
It also slightly reduces allocations and memory use and - at least on my system - improves benchmarking time for the whole system from 35.788s to 32.588s (but only one run each and 2 days apart...).

[codecov-commenter]

⚠️ Please install the to ensure uploads and comments are reliably processed by Codecov.

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 96.56%. Comparing base (b0b50d8) to head (c637f69).

❗ Your organization needs to install the Codecov GitHub app to enable full functionality.

Additional details and impacted files

@@           Coverage Diff           @@
##             main     #338   +/-   ##
=======================================
  Coverage   96.56%   96.56%           
=======================================
  Files          18       18           
  Lines        1863     1863           
=======================================
  Hits         1799     1799           
  Misses         36       36           
  Partials       28       28           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[timbray]

Looks good to me.
I wonder how I managed to not notice atomic.Pointer. Will leave a little while for pushing in case one of our other Go experts has an opinion.

[timbray]

BTW @arnehormann unless you object I will add you as contributor so you can run CI etc.

[arnehormann]

I don't object, but I don't know how often or how much I'll contribute after these PRs. Not that I won't, just that I don't know yet.
Thanks for your trust!

[timbray]

I don't object, but I don't know how often or how much I'll contribute after these PRs. Not that I won't, just that I don't know yet. Thanks for your trust!

Well, as long as you have a couple of PRs in flight, you should be able to trigger the CI/CD without waiting for me to wake up in West Coast time and press the "Run CI" button.

[arnehormann]

I don't object, but I don't know how often or how much I'll contribute after these PRs. Not that I won't, just that I don't know yet. Thanks for your trust!

Well, as long as you have a couple of PRs in flight, you should be able to trigger the CI/CD without waiting for me to wake up in West Coast time and press the "Run CI" button.

That... is indeed quite nice. I needed a lot of roundtrips until I broke no rules and a rather stupid blindness bug was squashed in numbits...

[timbray]

That... is indeed quite nice. I needed a lot of roundtrips until I broke no rules and a rather stupid blindness bug was squashed in numbits...

Our CI/CD setup is due to @embano1 and is quite effective, although as you'll see if you watch the ignore-case PR, it's got one problem that I'm having trouble working around.

[embano1]

Looks good to me. I wonder how I managed to not notice atomic.Pointer.

They were introduced somewhat "recently" in Go 1.19, and atomic.Value has been around for longer (and during the time you created Quamina) - so there's no-one to blame here ;)

+1 on using atomic.Pointer

[timbray]

Hey @arnehormann I have an overall performance test script and the change to atomic.Pointer led to an 0.52% improvement overall and 1.37% in the most common case.

[arnehormann]

That sounds wonderful! Is have hoped for more, but it's a lot for such a small change. Also, live by a thousand paper-mends or somesuch