One upvote is not enough.

I once wrote a commit message the length of a full blog post comparing 10 different alternatives for micro optimization, with benchmarks and more. The diff itself was ten lines. Shaved around 4% off the hot path (based on a sampling profiler that ran over the weekend).

Ew no.

Abusing language features like this (boolean expression short circuit) just makes it harder for other people to come and maintain your code.

The function does have opportunity for improvement by checking one thing at a time. This flattens the ifs and changes them into proper sentry clauses. It also opens the door to encapsulating their logic and refactoring this function into a proper validator that can return all the reasons a user is invalid.

Good code is not “elegant” code. It’s code that is simple and unsurprising and can be easily understood by a hungover fresh graduate new hire.

Yes. I’ll read the content, but I try to avoid interacting.

Mind you, db0 himself is a tankie, although he doesn’t seem to insist on imposing that on the users or communities on his instance.

EDIT: I stand corrected. Apologies to db0 for lumping him in with that crowd.

Most AIs are trained on older poster art like this - they’re well labelled, have consistent style, and because they’re older there are likely to be a bunch of duplicates in the training set.

Pretty sure this one predates AI art.

It’s from 1986

It’s Loss

None built in from what I recall. That was from back in 2011, so it’s possible things changed since.

Reading through, it looks like retries do exist, but remember that duplicate packets are treated as a window reset, so it’s possible that transmission succeeded but the ack was lost.

I remember the project demos from the course though - one team implemented some form of fast retry on two laptops and had one guy walk out and away. With regular wifi he didn’t even make it to the end of the hall before the video dropped out. With their custom stack he made it out of the building before it went.

I’ll need to dig through to find the name of what they did.

To be fair, because of window size management it only takes 1% packet loss to cause a catastrophic drop in speed.

Packet loss in TCP is only ever handled as a signal of extreme network congestion. It was never intended to go over a lossy link like wifi.

Only on signup

Anything using Blind as a “verified industry source” is going to be skewed to the type of person who uses Blind. Beyond that, it’s low sample size, and there are suspiciously round fractions for some of the larger companies. Worse, because Blind is blind - this doesn’t represent current employees, but merely people who worked at some point in the past at those companies.

Not saying it’s not good - just saying not to get overly excited over a badly done survey

That makes a lot of sense - I wonder if they also do the SIGSEGV trick like HotSpot to know when they need to JIT the next chunk of instructions

But does it run Doom? Using CMOV instructions only?

I thought FAT binaries don’t work like that - they included multiple instruction sets with a header pointing to the sections (68k, PPC, and x86)

Rosetta to the best of my understanding did something similar - but relied on some custom microcode support that isn’t rooted in ARM instructions. Do you have a link that explains a bit more in depth on how they did that?

From what I’ve understood of this - it’s transpiling the x86 code to ARM on the fly. I honestly would have thought it wasn’t possible but hearing that they’re doing it - it will be a monumental effort, but very feasible. The best part is that once they’ve gotten CRT and cdecl instructions working - actual application support won’t be far behind. The biggest challenge will likely be inserting memory barriers correctly - a spinlock implemented in x86 assembly is highly unlikely to work correctly without a lot of effort to recognize and transpile that specific structure as a whole.

Seriously though - JIRA isn’t always a massive pain in the ass. It’s just the way it’s used that sucks. Workflow restrictions so devs can’t move tickets from testing back to in progress, dozens of mandatory fields, etc.

When your tools start dictating your workflow rather than the other way around then it’s time to switch tools.

Friends don’t let friends use JIRA

I have either written or gotten a variant of every single one of these comments 🫠:

Please include the JIRA task in the commit title.

Did you run any manual testing?

Where’s the PRD link in the commit message?

Can you please split this into multiple smaller commits?

Can you combine these two commits?

Did you email Jon about this because he’s working on that project with Sarah and you might be duplicating efforts.

This should be named BarFoo instead of FooBar.

Why aren’t you using CorporateInternalLib16 that does 90% of this?

Why aren’t you using ThirdPartyPaidLibByExEmployee?

Why aren’t you using StandardLib thing you forgot existed?

All our I/O should be async.

All our hot loop code needs to be sync.

This will increase latency of NonCoreBusinessFlow by 0.01%. can you shave some time off so we can push in feature B also?

Please add a feature flag so we can do gradual rollout.

What operational levers does this have?

Lgtm - just address those comments

If you’re aware of someone or a company violating sanctions, you can earn up to 20% of the fines as a whistleblower.

Considering the fines start from 100,000 per transaction, you could easily retire on a single report.

I am disappointed this doesn’t begin or end with The Elements by Tom Lehrer

Holy propaganda batman!

The list of articles on that website is…extremely focused on one subject only.