I gave up counting but it did slow down.

Got one that I'm pretty sure is just... I didn't implement a feature and they re-framed it as a vulnerability. Okay dood.

different outage. (10th vs the 16th). Also that person with the popular thread is a nontechnical rage-baiter.

That said I had a great time with Pokopia last week. I experienced brain death.

Ding Ding Dopa Mine.

I really like the concept of the Evercade as I dream about being More Offline but mentally making the switch is probably too hard.

Similarly I wish the playdate had a GBA SP-like upgrade with a light and better ergo.

Lot easier to be less mad at games when you just give up on the new "big" stuff.

Brad was willing to help but it's been hard getting our schedules synced up and I'm not too useful for that work. Needs careful thinking over backcompat and idiomatic Go... Client won't need to change very much afterward.

I totally understand the original issue; First time I ran into internal port exhaustion was with ephemeral MySQL connections 25 years ago ;)

I'm not a Go person or getting the client fixed up has been hard. I'm willing to maintain if we can get the backlog dealt with.

re-reading this; should at least be using getmulti when batches are apparent, but there's a different issue with batch timeouts that you can run into (docs.memcached.org/advisories/g...) - if you haven't already thrown out mc let me know if it'd be helpful to fix up the client.

In more detail: you share a client object via some mechanism and the client batches disparate requests together opportunistically. Then responses get redirected back out later.

If that's scary you can still pipeline any kind of request together for the same logical client.

Yeah... proto goes beyond that. You can multiplex clients on few sockets. Some clients can, say: "use 16 sockets and stream req/resp's". Syscall batching can help overall perf + avoid socket gluts like you hit. @bradfitz.com -> we should try again to fix this? (we started a year ago and I fell off).

@calabro.io Just saw the outage writeup from the 10th! That looked rough. Which MC Go client are you folks using? The connection explosion is usually from clients that can't use protocol pipelining.

Reminds me of this: www.reddit.com/r/Android/co...

happened to a friend. dialed 911 and his phone locked up (recently!)

trying to envision a portable game console designed for factory sims.

Moved my old copy of Deep Work to a visible shelf in the office so I can have a constant reminder of how much I'm fucking up.

Just spent entirely too long reading all the replies here. As someone who's worked in or with dozens of actual DC's the discourse is wild. It's like watching Trekkies vs Star Wars arguments where neither side has watched either series.

You have so much patience.

Really don't want longer posts on here. Want people to formulate their weird opinions in blog posts then link them here at people all day. Update it if they want. "Yeah I could explain this to you but it's easier to act like you should know already" FFS, man.

okay okay: so we used to give security people bounties for finding bugs... but what if... security people pay maintainers instead? We could walk away and not fix it and then you don't get points or money. Pay us for the points.

Why are all these leeches after points anyway? What does points on their github account or name get them? A job?

Everyone's got their hand out...
"My AI agent found a bug, I need points (CVE/credit/etc)"
I can't afford to spend my time every week handing out points.

Are there only N bugs in the codebase? Do I need my own "agent loop" to find bugs before they do? Same time sink?

another one.

at least one more

What if I expose the event queue depth? the time-to-process event queue? How to communicate to user without confusing them since that's still not a measurement that says "a request takes this long"?

Keep trying to blog about a (effective) deficiency in a thing but getting distracted by trying to fix it.

Five more.

Five more just came in.

Am I a garbage programmer? Seems like a lot of evidence to this.

I don't want to fix them. I don't think I care.

please don't nuke me I just felt more empathy than usual to the detractors here. I've followed DF for a long time!

The vast majority of people aren't going to notice, especially if they have nothing to compare it with. PWM in the switch OLED bothered me more since I was sensitive to it; loved the screen otherwise!

Caveats: I'm older than dirt and have owned tons of peanut-butter class LCD's. I can see the issue. Given the 2's large high res VRR screen (and noting the other compromises the 2 has) I'm not sure I would change anything. An SD OLED costs $100 more (and is unavailable!)

[Can't possibly add enough caveats before this] - these videos did bug me when I saw them. Esp. the monitors unboxed thumbnail/opener. "Shame..." "bad screen?" to an owner who isn't going to watch the whole thing it didn't smell like a nuanced take. Other YT'ers bait so a casual user would dismiss.

Decided to become an F1 guy this season. I want to follow a sport but I hate most of them? I like engineering and zoomies. Also british commentary; basically no respect for anyone on the track, lol.

The level of "is this a security bug?" is so high now that I'm pretty sure I'm just gonna let them go to the public issue backlog from now on. Lets roll the dice.