uberduper

My point is that by letting the customer define the solution rather than explain the problem, nobody knows they're even trying to hold it upside down. The why of it all is lost in some PM / Engineer power struggle that usually results in ice cream cones having covers (and a happy customer).

> If a user holds an ice cream cone upside-down and their ice cream falls to the floor, do you blame the user for not holding their ice cream cone upright or the creator of the ice cream cone for a stupid design that allows the ice cream to so easily fall out of the ice cream holding device and onto the floor?

Playing along with this analogy, what I think we see a lot of in product development is the customer going to the PM and saying they need the cone to have a cover. The PM and the customer iterate over the specifics of the cover. PM goes to the engineer and tells them the cone needs a cover that meets x, y, and z requirements. Engineer, knowing how you're supposed to use the ice cream cone, objects. PM, knowing what the customer needs, insists.

I worked at a place that made a robo-calling sales crm and they thought it would be great if during an offsite they had everyone use it to cold call places and run through some script they'd prepared. They didn't seem to understand nor care about what sort of anxiety that could cause not-sales-people. That's when I decided to start looking for another job.

In my experience, if there's one thing the AI assistants do really well, it's understanding existing code and summarizing it or pinpointing specific things I'm looking for.

I'm not really a developer but I routinely need to go dig through codebases looking for answers and AI assistants have made this so much faster. This has been a boon for me.

> Or a more detailed disclosure:

> I consulted ChatGPT to understand the codebase but the solution was fully authored manually by myself.

What's the reasoning for needing to disclose this?

This is a chapter in a book targeting people working in the machine learning domain.

Kubernetes will handle this automatically if you configure the Topology Manager.

This is awful advice. Simple and elegant design does not start with dismissing potential problems.

Those questions are all prompts to have a discussion in lieu of tech trivia hour. Those responses do not demonstrate wisdom, they reveal a lack of maturity. It's not the interviewers fault you refuse to be interviewed.

I really like Nix package management. I really hate Nix configuration management. I already know how to configure everything. I don't want to have to figure out whatever abstracted config options some random package maintainer decided on.

The people I've talked to that have been using them say there's no setting for it and they don't hear anything like audio picked up on one side routed to the other earbud.

Bose made a product ~10 years back called Hearphones which were far more capable than what Apple is doing here.

IIRC Jabra earbuds have had "hearing aid" features for years. They, unfortunately, don't help with single side deafness the way the Hearphones do.

Apple isn't doing anything groundbreaking here but they are doing it at a very competitive price. The airpod features also do not help with single side deafness. :(

Frackers bought the Kermit Sand Dunes in Texas 7 or 8 years ago cause the sand there was ideal for fracking.

Rectangle and Spectacle before it are really great for managing window placement and sizing. I wouldn't use a mac without it.

But I don't use a mac anymore. I use linux desktops and there's just nothing quite like Rectangle. For a long while, I used a tweaked fork of pygrid. These days I use wayland compositors and don't have a good option. I've been meaning to hack something together to work with floating windows on Hyprland but I never get around to it.

Someday I'd like to write something that does placement and sizing like Rectangle but snaps adjacent window edges together. Like if I have a bottom left window that's 1/3rd screen width and assign a new window to bottom right 1/3rd width, the bottom left would get resized to 2/3rd wide to fill the void. If I place a new window at bottom middle 1/3rd, the bottom left would get resized back to 1/3rd. If I resized bottom middle to 1/2 width, then the bottom left and right would each be resized to 1/4 width.

Hyprland has a plugin system that I believe would let me do this (someone has created an i3-like tiling plugin) but I don't have the skills or motivation to learn c++.

I'm reading through the bug report and the investigation in the comments. Is there something here that makes this bug extremely obvious to you that nobody else is comprehending?

pmtu

If you're in the US, talk to your GP, family Dr or w/e. They can prescribe but they cannot diagnose. They'll refer you to someone. You probably want to be referred to a counselor rather than a psychologist.

You should _really_ consider therapy in addition to any medication. The meds will be amazing for a month and then useful for a year. After that you're going to have to make a decision about how you want to continue treatment. Keep bumping the dose and adjust to the side effects or 'typical therapy'. By then, that therapy is going to be much harder to apply to your life than it would have been in the first few months when you started the meds.

In my experience and observations, calorie restricted diets are pretty easy to adhere to until sometime between dinner and bed time. This is where people typically go off the rails and eat far in excess of satiation, negating any success of the day.

Intermittent fasting typically has people shift their eating times towards the end of the day which should significantly reduce the chances of falling into a pre-bedtime eating habit. Especially if you also shift your sleep window earlier. So if you stop eating at 8PM and make sure you're asleep by 10PM, you're less likely to get hunger pangs.

It's not usually the battery backup or the generator that fails. It's usually the switching equipment that has to go from mains to battery to generator to battery to mains. And doing it without causing a voltage sag on the generator.

I lost track of how many datacenter outages we caused testing the power backup/failover back at eBay in the mid-2000s.

There's no winning when it comes to power redundancy systems.

I do it today with 1password. I did it back then with yodlee and mint.

I've always avoided including any email, phone or utilities account creds since these are, unfortunately, ways financial companies frequently use to authenticate you or authorize changes. I didn't want to be locked out of them in the event my bank or trade accounts got taken over.

Mint was originally built on the back of Yodlee.com which is something I had a small part in building. I was gone before that deal was.. minted, so I have no idea what the terms were like. My options ended up being worth like $15 and Mint got bought by Intuit, so I'm guessing not great for yodlee.

AFAIK they only used yodlee for the first couple years until they were able to secure their own connections to financial institutions.

Makes me a little sad thinking about how long I've been at this now..

It's common for performance based RSU grants to stack up to this sort of total comp over a few years.

That's not at all what I meant.

I've been half expecting Steam (or someone) to create a cross-platform runtime/vm for game developers and hardware vendors to target.

Okay that makes sense. Thanks.

How quickly does a glucagon shot trigger GNG and how quickly can GNG raise glucose levels?

I may be misunderstanding something here. I'm assuming the emergency you're referring to is low glucose. Why would you use a shot of glucagon rather than a dextrose tablet for that sort of emergency?

This is my primary objection to 3rd party kubernetes operators. I really don't care how hard something is to setup or learn to manage. I care how hard it is to recover, tune, and scale.

Sure. SLE could become (the basis for) OpenELA10?

Seems to me that all the RHEL derivatives were perfectly content to let RH set the standards for enterprise linux. RH doesn't like that arrangement. SUSE probably wouldn't mind it.

All the derivatives have their EL8 and EL9 based distributions they've committed themselves to support for some time, so I'm sure they'll maintain bug for bug compatibility as best they can given the circumstances. No reason for them to keep following RH after that.

Seems like an obvious next step is to get enterprise hardware and software vendors to begin certifying on OpenELA in addition to RHEL. OpenELA10 or w/e could stand on its own as a competing enterprise linux standard that everyone is free to build from.