you can have wpa2 and wpa3 on the same radios.

IME this caused all clients to disconnect and cease being able to auth. I just don’t run mixed auth for that reason.

2.4Ghz/W.G has WPA2 and 5Ghz/W.N/W.AC has WPA3. That’s what works for me on my LinkSys WRT 3200AC.

They still make use of desktop CPUs that can pull 120W when idling and up to 250W under heavy loads. So light on power? Maybe. But power sippy? Not against an M4 Mac, which can be as little as ⅙ as much under a similarly heavy load.

Yeah, but this is Rust, not some single-dev pet project. You would expect a fully fleshed out framework to have accessibility features built in as first-class features, not some randomly bolted-on afterthought.

I already successfully block all Reddit ads. I use Old Reddit exclusively, as I cannot stand New Reddit.

So if Old Reddit goes, so do I.

IIRC you can attached scanned/photographed content to his profile entry. So at least it’s saved somewhere.

If she happens to have an entry on Ancestry, the person maintaining that record might want to scan that info in and attach it to that entry.

Fun! I’ll get to moderate a Discord server for blind people that blind people can’t use on desktop!

How would a rewrite cause this to happen? Is there something about Rust frameworks for the desktop that exclude accessibility features?

I’ve been working primarily with C#, but none of their frameworks exclude accessibility features… would be shocked AF if Rust frameworks just left that all out.

Looks like the RAM crunch is starting to affect more than just greenfield projects.

Good. Electron is a bloody memory hog. We don’t need to squander 300Mb of memory just to display 15Kb of text. And then rinse and repeat with multiple other apps like BitWarden Desktop.

You listening, BitWarden? The browser add-ins are fine, the full-fat desktop app needs to lose some serious weight.

Every technology you touch hurts the environment

All life causes pain and suffering to other life.

Therefore, the logical end-stage conclusion is to trigger a false vacuum decay event that erases all of existence throughout the universe and so relieves all suffering.

no confidential client information is to be sent to any cloud AI service

If the code has been crafted correctly, none should be.

Client info should be 100% data, and remain 100% in production. If you have any dev-side data to test and debug with, it should be 100% procedurally-generated pseudo-data that only imitates client data in terms of gross patterns, and not specific points of data.

Like customer names. Those names should be broadly representative of what is found in the regions the software operates in, in order to demonstrate acceptable data input validation and effective storage compatibility with those names. For example, you don’t want to be storing European names in a DB field that is not UTF-8 compliant, or validating those names by restricting characters to US-ASCII. But any one generated name that just so happens to match an IRL name should be a statistical occurance, and any extended data attached to that entry should (generally) not match any other IRL data for any IRL occurance of that name…

Why should I pay a subscription fee? The AI is not touching my code, not generating any of it. I use the free versions to ask questions or give it snippets as examples, and go from there.

Sure, I’m beginning to bump into Claude’s free ceiling, but there are so many other options out there that I just partition my questions to the most appropriate model for that question.

There may come a time when I actually sign up for a paid tier, but it hasn’t happened yet. Until then, I will continue to exploit the parasites that fund this largesse.

And honestly, I won’t be upset if it all crashes and burns – it’s only a tool for me, not a crutch like with so many other devs as of late. Almost all of them will have regressed all the way back to junior-level skills if they had to hand-code all over again. It’s actually kind of sad to see such skills erosion and dumbing down that’s happening IRL.

Decades ago we had engineers and devs rise up through the ranks to become a part of steering committees.

Now they’re all MBA/BPM/PMBOK people with either zero or superficial tech/dev knowledge.

You cannot “tightly scope” that which you cannot understand.

Bottom-up should be the direction.

But- but- but- that’s communism.

Workers taking control of the means of production is heresy in capitalism.

The one thing that rubbed me the wrong way is that it treats the underlying host as “disposable”, in that it intentionally does not support in-place OS upgrades. As in, if you want to jump from an older LTS to the most recent, you need to explicitly spin up a blank, new LTS host, run the installation script from scratch, and then restore the data from an overnight backup.

I can understand an immutable container/binary with external data storage and config settings, but flipping the script and locking down the underlying OS? Yeah, that seems excessively unnecessary.

IMO, using AI to advise or guide you is the way to go, as it’s not actually touching the code.

But AI doing the work is a disaster that will lead to epic amounts of technical debt because it is impossible for AI to utilize judgement or taste.

I have been using AI more and more lately to just advise, to find, to make me aware of other options. I still do 100% of the work. So while AI hasn’t made me faster, it has made me more flexible by kicking me out of ruts and making me aware of alternative paths to reaching an objective.

I will not let AI touch my code until/unless it acquires judgement and taste. But until then, it’s turning out to be one hell of a gopher and rubber ducky.

then you end up on a VPS with a bad spam/RBL reputation because they have other customers being spammy dickheads, and their entire IP allocation is listed.

Sure, if you don’t do your own research and legwork, first.

In fact, there are a shitton more bad corners you can paint yourself into if you are too fucking lazy to do some trivially basic research ahead of time, from hosts to software chosen to how you configure things.

Hell, a lot of these bad corners also involve turnkey solutions hosted by others. So simply having someone else do all the work can hose you just as badly. Sure, Azure/O365 or AWS Workmail or Google Workspace solutions are usually not that kind of corner, but unless you are willing to pay out the nose for minimal needs, you’re usually going to go for something a lot smaller and possibly more sketchy.

Even in this day and age, when furniture is being mass-produced, there is still a niche for those who create furniture by hand, without the use of assembly lines and CNC.

The same goes for software development.

The downside is that you need to become the creative, the “ideas man” that can come up with that bespoke solution that AI can never truly provide. And you likely have to do it on your own, or if you are lucky, in a small company whose entire identity relies on bespoke solutions and high-quality, high-cost work.

The vast majority of software developers will never be able to make that jump, just like how hundreds of thousands of furniture-makers in the early 20th century, across North America, were also unable to make that jump. Those turning out bespoke furniture today represent an almost-unmeasurable rounding error compared to the same numbers of craftsman a century ago.

But they still exist. They can still make a living. And so can software developers with the right skills and attitudes.

Can’t fleece us enough with stupidly-expensive junk food, now they’re also cramming advertising down our howling pie hole

the planet will live on and heal.

Unfortunately, this line of thinking ignores the elephant in the room: inertia.

There have been prior climate change events that have gotten up into the +8-12℃ range (which, BTW, means palm trees at the poles and the extinction of any land species over 45Kg/100lbs), but the difference there is that they have taken many hundreds of thousands of years to get there, allowing whole ecosystems to migrate across entire continents while continuing to sequester carbon.

Our changes to date have - when compared to the very-next-fastest event in history - compressed over 150,000 years of climate changes into less than 75.

This means that not only is there a massive amount of inertia involved with our changes - several orders of magnitude greater, which will force CC to continue on well after humanity completely ceases adding to the problem - but ecosystems are just going to die in place. They simply aren’t being given the time to migrate or adapt, leading to significantly unoptimal/degraded CO2 sequestering, compounding the problem.

Or, in other words, this inertia is destroying the brakes that could slow and eventually reverse CC.

Finally, as the sun ages over the millennia, it throws off more and more heat. The global average needed to cause a runaway Venus Scenario has slowly been creeping down over the last few billion years, and may have already crossed the +12℃ level.

Which means that, even if humanity just ceases to exist today, there may be enough inertia to kick off sufficient feedback loops (methyl hydrate eruptions, permafrost eruptions, etc.) to push the planet into that Venus Scenario, extinguishing all life on Earth.

Sweet dreams.

Source: BSc in Physics & Math, former classmates in the CC & biology fields. I read the pre-print evidence they are collecting and can see where it points.

Don’t sign in with Google or Apple or Facebook. They get data out of it and lock you into their platform.

This is why I have never used some services such as PushBullet. I mean, this service would be absolutely useful to me, but without a direct user login it’s persona non grata.