Justin Nuß

Programming enthusiast and Gopher

@[email protected]
@nussjustin.bsky.social

[Justin Nuß @nussjustin]

@mr_wormhole @valyala @_rsc But if it does not that could confuse all the people who expect it to, since it always did so, and having reproducible output can be very useful. Also the comment in the code says it should always give the same result and that was written by Russ, so yeah.

[Justin Nuß @nussjustin]

@valyala @_rsc I know but the playground is supposed to be completely deterministic so that output does not change and can be cached. At least that's how it was in the past and the playground even faked time and sleeps in order to enable this.

[Justin Nuß @nussjustin]

Reminds me of the time I found and optimized a query that took took 2 hours to generate some kind of product export (and which had to run twice)

Got it down to <1 second with 0 changes to tables/indexes

Still not sure whether to love or hate SQL

[Justin Nuß @nussjustin]

Sometimes I'm not sure whether I love #sql or hate it

Just optimized a SELECT that took ~2 hours by moving some JOINs and subselects into CTEs

Same data returned, but optimized query takes < 1 sec

#mysql #performance

[Justin Nuß @nussjustin]

There goes my Saturday, but I finally finished the initial rewrite of my Go resp package for RESP3. Still needs more integration tests. Next step: Trying to integrate it into an existing Redis client package. https://t.co/d4UfLJY1C7 #golang #redis

[Justin Nuß @nussjustin]

Wrote myself a userscript (Tampermonkey) that shows links to the relevant CLs on the #golang release notes. Makes finding the relevant CLs much less annoying (no more dev tools or source viewing) https://t.co/Nk0gu2krdg

[Justin Nuß @nussjustin]

Reminder: Just because SSDs are fast doesn't mean you shouldn't optimize your application's I/O.

Just went from reading ~800MiB/sec to ~2200MiB/sec by switching to direct I/O, using larger buffers, and doing concurrent reads.

[Justin Nuß @nussjustin]

@psjarlo @pmjordan @kellabyte @janl Cloudflare has a blog post on this (in their case it was impacting ChaCha20-Poly1305 performance): https://t.co/Khk0czDP9s

[Justin Nuß @nussjustin]

@peterbourgon using time.AfterFunc would be more efficient and shorter here https://t.co/04MMg2jNoJ

[Justin Nuß @nussjustin]

@peterbourgon Aaaaaand there's a race in the tests. Darn. Just sent a PR to fix the tests

[Justin Nuß @nussjustin]

@peterbourgon If you do flag.Parse before flag.Set you can even prevent `go test -run .` from running any tests

[Justin Nuß @nussjustin]

@peterbourgon func TestMain(m *testing.M) {
if runtime.GOOS != "linux" {
flag.Set("https://t.co/GjQJPvgZaB", "^$")
}

os.Exit(https://t.co/jT1dWMv3n3())
}

[Justin Nuß @nussjustin]

@peterbourgon Now I'm even more scared :-( Here, take this and let me live! https://t.co/q9pMiahYck

[Justin Nuß @nussjustin]

Talking about adding global state to the stdlib, I always feel like @peterbourgon is watching from somewhere.

[Justin Nuß @nussjustin]

I hate Mondays

[Justin Nuß @nussjustin]

TIL: Ab 12 Kugeln gibt es im Eis Cafe in Hude 2 Becher. Hups. #themoreyouknow

[Justin Nuß @nussjustin]

@mholt6 @caddyserver You never change a running system. Unless httpd segfaults (again), then change it to one process per request.

[Justin Nuß @nussjustin]

I've finally found the perfect use-case for desktop notifications. Displaying random #golang proverbs! Why didn't I think of this earlier?

[Justin Nuß @nussjustin]

my go exercise of the day: implementing readv for net.Buffers based on the writev logic from @bradfitz; lost much time on obivous errors