> I was hoping that there would be a "solution" of sorts to tackle / handle this issue of when EVERYBODY seems to use this strategy, but perhaps there isn't one...?
We're overdue for a major war, which will be reset on how we treat other humans by the end of it. Humans killing humans on an industrial scale between near-peers is followed by periods where people realize that maybe being dicks to each other isn't the ideal state. More cantankerous politicians being elected only increases the odds of war breaking out due to diplomatic failures.
Obviously gp forgot that open source projects fall under the ambit of open source law, witten by open source legislators and adjudicated by open source court circuits /s
The "west" absolutely has a monolithic IP jurisprudence, rules codified by trade agreements and passed laws[1]. This has harmed open source in the past, such as the multi-decade, recurring issue where folk have the license to make use of the source code, but not the associated trademarks.
1. As evidence, I offer the great mobile patent wars of the 2010's, which played out similarly across multiple countries the suits were filed, except the home-countries where the scales were the scales were thumbed.
My hot-take: this is the future of high-fidelity prompt-based image generation, and not diffusion models. Cycles (or any other physically based renderer) is superior to diffusion models because it is not probabilistic, so scene generation via LLM before handing to off to a tool leads to superior results, IMO - at least for "realistic" outputs.
Of course no one knows the future, but I think it's very plausible that the future of films/games (especially games) tech resembles something like this:
1. Generation something that looks good in 2D latent space
2. Generation 3D representation from 2D
3. Next time the same scene is shown on the screen, reuse information from step 2 to guide step 1
I made no such claim. The only thing I declared is my belief in the superiority of PBR over diffusion models for a specific subset of image-generation tasks.
I also clearly framed this as my opinion, you are free to have yours.
> also clearly framed this as my opinion, you are free to have yours.
Yes, thank you for your generosity.
You very clearly framed your opinion as a dual choice ("this instead of that", "rather than", "either or"). The most natural way to read your comment is that way: one or the other.
That's the way the English language works. If you meant something else, you should have said something else.
You may have missed my first 3 words that did the level-setting. Infact, the first word should have been enough to signal subjectivity, that's just how the english language works.
Hot take: a piece of commentary, typically produced quickly in response to a recent event, whose primary purpose is to attract attention
> And I suggested comparing releases, not commits...
Who's asking for all those new features? Enterprise users? I love the old Gitea because it was light enough to run on a Raspberry Pi Zero. I've been delaying trying Forgejo due to concerns about bloat, but you've just sold it to me. All I need for it is to mirror Git repositories I care about every couple of days. Like gp, I consider Gogs/Gitea to have been feature-complete years ago
Aider (cli) and continue.dev (VS Code plugin) can both run with a local(net) Ollama. The qwen-coder models are pretty good and getting better; qwen3-coder is in the ballpark of Sonnet 3.5 for code-synthesis, albeit slower on my hardware.
> we don't like to provide those (both payers and receivers prefer not to give the other participant their ach numbers
This is because in the US, anyone can pull money out of your account with only the ACH numbers; which is an insane design[1]. In most other countries, the worst you can do is deposit money. The equivalent of ACH pulls requires significantly more paperwork and proof of consent by account owner.
1. Much like SSNs, which can be debilitating if not kept secret. US payments run in "true names" magic, and simultaneously expect you to register with your one true name at random places with questionable security practices, and it's your fault if there's a breach.
> This is because in the US, anyone can pull money out of your account with only the ACH numbers; which is an insane design
That’s the default for at least Germany and SWIFT, too. You can ask your bank to disable this, but that means losing the pull functionality completely; I think some banks have an interface to whitelist individually, but the majority doesn’t.
It can become a problem especially when you list your account number publicly somewhere for payments or donations: somebody will eventually use that account number to pay for random stuff. You’re contractually obliged to check your bank statements and ask for a (free) chargeback within a certain period of time (some weeks?).
At our projects, we solve this by having a separate “public“ bank account for incoming donations that blocks pulls, and a much less public one for pulls.
Apart from this use case, abuse seems to be rare enough that banks typically don’t expose the functionality to disable but only do it manually when asked specifically. I doubt most people even know they could.
> In most other countries, the worst you can do is deposit money.
So in the EU, anyone can indeed pull money with your account number (and with RTP that may change someday). But we can also revoke any such direct debit within a certain period of time.
I had to do it once, over my banking app, money was back the moment after I clicked.
The "anyone can pull money out of your account" piece is true but it also isn't.
Yes, if a financial institution allows you to originate a debit from another account without verification, you could take money from anyone's account. The max liability you should have given prompt reporting of fraud (less than 60 days) is $50, and if your institution doesn't give it all back then find a new one.
ACH is also technically reversible, whereas wires/other instant transfers are not.
FIs also do fraud checks on ACH, I believe it may be a regulatory requirement now (sometime in the past few years?) to have some form of fraud check before sending originations to the FED. Typically this is verification of the other party being a known entity/account, which would ideally burn fraudsters very quickly.
Most transaction facilitators don't play around with any of this though, and have some "account linking" step before they are willing to originate transactions. Micro-deposits that you would need to verify on the other account.
> This is because in the US, anyone can pull money out of your account with only the ACH numbers
Whoa, I don't blame people for not wanting to provide ACH numbers in that case. Is there any groundswell to provide a system where this doesn't happen?
FedNow is that system. It only sends money. Can request money but it needs to be approved.
However, I'm not sure if it uses different account numbers from ACH. It could be that sharing account numbers could be secure with FedNow and dangerous for ACH.
No, since software if built around existing limitations. IE stripe invoices will generate a single use account number per invoice, so you can give that to someone who needs to pay. Aligns payment with invoice since they’re 1:1 and you can pull out funds since it’s always empty
For Linux desktop gaming, Steam bests GoG all day, everyday.
Before Valve sponsoring/partnering with Code Weavers on Proton, running anything-but-old-and-stable games via Wine was a fraught affair, now even games that update weekly/monthly run perfectly, without having to fiddle with config files or downloading specific DLLs. For the large and growing library of supported games, Steam made Linux gaming painless.
GOG will sell you Linux versions of games. In fact it doesn't care - once you own it you can download all of the game files for all the platforms it was released for. However, it only sells you what actually exists. It doesn't take the Windows version and run it in a Windows emulation layer like Steam has. If you want to run the Windows version emulated on Linux, you're on your own.
Not on your own; you're with a community that supports running GOG games on that Windows emulation layer. Heroic can run most GOG games very well. Steam is more reliable because it's Valve itself that's behind it, but Heroic is not that far behind.
> The real story here is that IP ownership is capital-intensive when it shouldn't be
The backbone of the US economy are services and software, which depend a lot on IP. Deliberately or not, "low-value" American manufacturing was sacrificed for these high-margin industries[1]. AFAICT, it's impossible to turn back the clock on manufacturing without disadvantaging US software/services both on the legal regime and trade fronts
1. Which is why SWE salaries are higher in the US that RoW. I don't think trading high-salary service jobs for low-paying manufacturing is a good decision, but lots of people - including the current executive - think they can get it all. My working theory is Europe and China are not dumb and without agency and are just biding time for decoupling, should their manufacturing industries be undermined by US policy.
We're overdue for a major war, which will be reset on how we treat other humans by the end of it. Humans killing humans on an industrial scale between near-peers is followed by periods where people realize that maybe being dicks to each other isn't the ideal state. More cantankerous politicians being elected only increases the odds of war breaking out due to diplomatic failures.
reply