Patrick Westerhoff

Crucial is shutting down — because Micron wants to sell its RAM and SSDs to AI companies instead. It’s already getting hard to build an affordable PC, and the exit of the longstanding provider of consumer memory is going to make that even more challenging theverge.com/news/837594/cr…

@lesbiansquiddo @TheOrbPonderer7 @CrocutaFang I will forgive you.

@TheOrbPonderer7 @poke @CrocutaFang FUCK.. we're so sorry man

why is the quag ep not out yet..... fuck my fucking chungus life @quag die die die

@lesbiansquiddo @TheOrbPonderer7 @CrocutaFang I usually am but you didn't ask nicely 😔

@TheOrbPonderer7 @CrocutaFang @poke i know right.... wow patrick you seemed like a nice guy 💔

@dariogriffo @vyrotek You could just target net8 in your libraries and be done with it. You can then consume them from both net8 and net10 applications. The target framework does not need to match the version you want to run it with. And you also do not need to update the target version all the time.

having to maintain and upgrade for example AWS lambdas, who are runtime dependant, new libraries targeting framework 10, but I want to use them in my app that is only 1 year old and uses net8 and can't because they dont target netstandard, which by the way, nobody talks about that anymore. All of this makes evolving so difficult, when versioning should be simpler: break something: increase major. add feature: increase minor. fix bug: increase patch. release often. Look at Go, they are on 1.xx for more than 10 years, literally means you can run anything and upgrade everything smoothly, they ship constantly minimising pain. Also this means that MS guys will only maintain 1 version, the last one. Obviously their mindset has to change and align with the whole SWE industry....

I never thought about this, but #dotnet MUST stop creating different versions every year. Unless something is fully broken Microsoft needs to have a framework that will work now and in 10 years This yearly release makes no sense. New features? Integrate them early and increase the minor version.

@Aaronontheweb What's up with those comments?! “for us it does not matter whether it is free, paid or who maintains it, as long as it is actively maintained” “The problem is that you're focused only on the negatives” Wtf, let your wallet speak and support the projects you're depending on.

Dude, you cannot win as an OSS maintainer sometimes - Matthias decides to step away from NUKE - doesn't change the license or anything, just decides it's not a great use of his time any longer - still catches shit in the comments

@eduardog3000 @nsg650 TrID has been doing this very successfully for ages… mark0.net/soft-trid-e.ht…

@nsg650 Except most (non-text) files can be identified by their headers. And even if you strip out the headers, most files have well defined formats that can detected without AI. A PNG without the 8 byte magic number (or even the IHDR chunk) will still have IDAT chunks.

Hard to believe there are some people who actually read/watches this guys journalism when this is his technical knowledge.

@Aaronontheweb @chriscanal Aspire is using Blazor Server? That is actually comforting (as someone who advocates for Blazor Server for internal applications a lot)

@chriscanal yes, this is the only thing that makes me nervous about using it. If .NET Aspire gets rewritten to use React I'm bailing out immediately - not waiting around to get Silverlight'd

I actually like Blazor Server because it's 1000x less work than getting SignalR + JS bindings equipped and can actually be type-checked using a single build system. I just wouldn't build anything that needs to be Google-indexed with it.

@apxltd @Aaronontheweb So you would rather depend on free and unsustainable solutions, or on self-made replacements built by amateurs that have no intention of maintaining it for the long run, than depending on a library that has a multi-year track record and now also a plan for a sustainable future?

@Aaronontheweb No one is looking at it this way. Instead, they are using a fraction of functionality of MT. They don't need to build a replacement, they just need something that fits their usecase. Removing a dependency on an unstable third party (FOSS, then 4-5 figures, and then?) is prudent

"So you're mad that MassTransit is going to cost 4-5 figures going forward" "That's correct, Dave" "And so your team of 8 is rewriting all messaging using the raw RabbitMQ drivers?" "Correct" "At an estimated salary/time expense of $950,000 in upfront dev cost?" "Correct!"

@apxltd @Aaronontheweb Open source libraries are used a ton in enterprise applications, in both simple situations and also very complex setups. I know multiple projects at the company I work at that used most features of MT, where they are now frantically looking for a (free) alternative…

@Aaronontheweb No doubt! But when they start as a FOSS implementation and community use, most organizations aren't using it in that capacity. It's just a logging library or whatever that someone added. The ones that get real value likely are already paying for support.

@peen420_ @MikeCodesDotNET @Aaronontheweb And the reality often is more like “I don't want my employer to pay for this thing”

@MikeCodesDotNET @Aaronontheweb "I don't want to pay for this thing that is cheaper than building a solution myself" is such a strange complaint in the first place.

Ah cool, so I suppose .NET users are going to run out and start contributing to / donating to projects with FOSS licenses - right? .... Right?

@Aaronontheweb Looking forward to seeing this person go a commercial license route in ~5 years after these forks gained any kind of relevant traction… (unrealistically assuming that they stay relevant and maintained for that long)

I wrote a comic about AI art theoatmeal.com/comics/ai_art

@Aaronontheweb Looking at how the EF Core team is evolving, I am happy for MS not to take on new subprojects that could further jeopardize the whole ecosystem. Better focus on keeping the fundamentals alive…

The built-in "Eventing Framework" issue in dotnet/aspnetcore was closed last night as "not planned." This is a tragedy for the least capable users in our ecosystem, having to resort to *shudder* evaluating mature third party frameworks to do the same thing.

Where do I start. OOP is not a discipline, it's a paradigm, a mode of thinking about how programming ought to be. OOP is not primarily about data structures. Not at all. An object is mostly an abstract interface. It might have some data, or it might not. Many OOP practices encourage creating objects that only hold references to other objects and no concrete data at all. The problem with programming is that computers can only ever move data around and perform some arithmetic operations on them. That's really all they do. Everything else is built on top of that. Display, audio, and networking at just I/O operations, and I/O to the CPU just looks like reading/writing to and from memory. There are levels of abstractions, and often times it's beneficial to isolate lower level concerns from higher level concerns. You can do that just fine with good old procedures. Sometimes, it's beneficial to have "abstract" objects that could potentially have many different implementations. A file system is the thing that comes to mind most of the time. I can read/write to files without worrying about the underlying file system. That's a good thing. Not everything is a file system though! Not everything needs to be abstracted and hidden. The criticism against OOP is usually around the level of granularity. Within your own code base, you want to know how everything works. You don't want to _hide_ things from yourself. It might be useful to limit the scope of visibility for some fields, but it's always on an "it depends" basis. OOP wants you to make strong boundary modules on very small scales, and it comes with absolute rules to help enforce that, like always making everything private by default, and always dealing with abstract objects instead of concrete data and procedures. When the QT below says that OO is a discipline, they mean that following the OO heuristics is always the right thing to do and it's better to just always do it and get used to doing it, and you don't get to argue about it, just like you always fasten your seat belt, always take a daily shower, always brush your teeth before bed, always go to the gym on schedule, always follow the correct diet. Don't second guess yourself every other step. Just do the right things and you will get the good results. The problem of course is that you do not get the good results from following the OO rules because they are wrong when applied as such. Some things can be useful all the time and you should do them even if they bother you a little bit, like fastening your seat belt. Some things are useful some of the time but not other times, like taking the highway vs the regular road, riding the train vs calling a taxi. The answer is always "it depends". This is not to say that there are no heuristics. There are, but not everyone agrees on them. Different groups of programmers have different heuristics. Some of them overlap, some of them don't. Some of my heuristics that I found to be immensely helpful in making programs easier to understand and reason about: - Prefer regular data and procedures - Prefer languages where structs have value semantics (as opposed to reference semantics) - Zero value initialization by default Everything else is "it depends". How big should the file be? It depends. How long should a function be? It depends. How big should a module be? It depends.

@GergelyOrosz Because in any negotiation, you have no power at all if you're unwilling to walk away from the table. If you have no deal-breaker, then you'll take whatever is offered, even if it's 37 cents per audiobook sale, which is, according to my calculations, what I would actually get.

@GergelyOrosz No one is coming to help authors. We have to help ourselves. And that means being willing to walk away from the table. Even if we miss out on 65-80% of the market.

How is it allowed that: - Audible has 65% market share in the US for audibooks - It offers 20% royalty share to authors if audibooks are sold non-exclusive (so, outside Audible as well). So for a $10 audiobook: Amazon takes $8, publisher gets $2! 80% take rate!!

It's not harmless. It's not "just an innocent thing" to be fooled by AI. You NEED to be able to discern and differentiate reality. You NEED to know how the world works on a basic level. Because fooling you with baby animals and flowers is just the start of something much worse.

The European Council has taken a proposal to force mandatory scanning of all photos and videos sent through private messengers (including encrypted messengers like Signal) and they’ve rebranded it as “upload moderation.” The implication is that it’s voluntary when it’s not.