mubou

Important part is at the bottom (really lazy journalism):

> The measure defines a deepfake as any video or audio recording or image that appears to a reasonable person to realistically depict someone doing something they did not actually do.

Interesting use of formaction on a button. One suggestion: put role="link" on it so it appears as a link to screen readers as well.

Unfortunately, by using a button you can't right click to copy the URL or open in a new tab/window. Middle clicking doesn't work either, although that could be fixed with JS. The URL also doesn't show up in the bottom left when hovering.

Edit: One thing you could do if you don't mind making your site JS-only is intercept link clicks and trigger a form POST instead, so the links are all normal links that respect default browser behavior, but the requests are all made as POSTs. Though honestly it's a losing battle, trying to stop bots; I've largely given up. You also have to consider what happens when someone shares the URL or bookmarks it, or tried to reload the page.

Looks like they're trying to do this: https://llmstxt.org/

I don't see this catching on any more than humans.txt did, tbh. The spec has some issues anyway: requiring .txt even though it's markdown, and requiring a particular URL structure for no good reason instead of leveraging <link rel="alternate">, which is already an existing web standard (i.e. there's no need for this spec in the first place). Putting the file in the site root instead of .well-known doesn't help, either; humans.txt did the same, but only because it was a play on robots.txt. It's not 2011 anymore.

"You didn't register your trademark, so I guess you don't care if I use it!" Setting aside mermaid here, that line of thinking is not only legally precarious but heavily unethical. In the US at least, trademarks are implicit, like copyright. Registering a trademark can help in court, but it's not required.

Well, at least they're supporting the project and they're ok with it.

It's not uncommon for companies to piggyback off an open source project's success and mislead people into thinking it's theirs. Seemed an awful lot like that was the case here.

I've never understood the popularity of McDonald's in Japan. They don't have quarter pounders (only Big Macs + the cheaper cheeseburger menu items from the US menu), and you can't get a burger with "the works" or "deluxe" either (you can ask for tomatoes at the counter, but it's charged extra per slice, and some staff don't even know it's an option). It's a far cry from its American counterpart.

(Weirdly, they do actually have the quarter pounder patties, but only for the soy sauce burgers.)

Was expecting the common "background-image: data url + filter: blur" that a lot of static site generators produce, not a binary encoding algorithm implemented in CSS! Very impressive.

I wonder what other things could be encoded this way. Those generic profile pictures, perhaps? (The ones where your email or account id is hashed to produce some unique geometric pattern.)

The problem is that a lot of people will google mermaid chart, or come across an extension like this, and find a site that looks like the official website and docs but isn't; it's a paid product with no relation that's simply profiting off of the name. That's deceptive, regardless of whether it's "fair use" or not.

Honestly, I'm surprised anyone on HN would defend that sort of behavior.

It's worth pointing out that mermaidchart.com is NOT the official Mermaid project (https://mermaid.js.org/).

They're a for-profit company using the Mermaid name.

This creates a lot of confusion whenever the name comes up, which I suspect is deliberate.

Curious how far they intend to take the "email hosting service" side. Will it be limited to e.g. @thundermail.com domains or similar, same as Gmail & Outlook, or will I be able to bring my own domain?

And if we can BYOD, will they let us have a catch-all? I want to be able to sign up for a service with company@mydomain, so if I get spammed I know who sold my data / got hacked and didn't admit it. I think Google Workspace and Proton are the only major email providers that support that. But the former means switching to a business account which doesn't always play nice with Google's other services, and the latter doesn't work well with other email clients (by design, of course, but I'd be worried about lock-in as a result).

Regardless, I have reservations about trusting Mozilla with webmail. The "Looking Glass" fiasco from years back showed they don't mind selling out to brands, and their current financial trouble might make them more likely to do so again. I realize the Firefox fans will downvote me for saying that, but it's a legitimate concern.

Are you using `shasum -a 256`? It defaults to SHA-1 otherwise.

I get the correct hash on linux using sha256sum.

Edit: You may need to pass -b (--binary). The man page[0] says it defaults to "text mode," whatever that means (this SO question[1] for md5sum suggests it has to do with how it opens the file, which apparently makes no difference on Linux/Windows but maybe it does on Mac?)

[0]: https://ss64.com/mac/shasum.html [1]: https://unix.stackexchange.com/q/127959

A bank replacing their engineering dept with AI is all sorts of wrong. What do you want to bet in a few months they'll be scrambling to hire devs after some AI screwup lost them millions of $$.

That unnecessary intro before the homepage is displayed reminded me of those awful flash intros sites had back around 1999–2003.

(I tried to find an example but they're all broken now since Flash is dead.)

I agree. It's far too easy to make a change to your model that removes/renames some property, but not update one template that you forgot uses it too. Without screendiff testing, that sort of bug will easily make it into prod.

This is one of the reasons I like C#'s .cshtml (Razor) syntax. The entire file is compiled to C#, so when you do `<div>@Model.Foo</div>` the build will fail if Foo doesn't exist. String-based (rather than compiled) view templating is, IMO, a mistake.

> Google's JS style guide also focuses on JSDoc for the same reasons

Google uses TS internally. Any JS is strictly legacy code at this point. The link you posted even says this at the top:

> Please note: This guide is no longer being updated. Google recommends migrating to TypeScript, and following the TypeScript guide.

Also, Google's style guides are not really authoritative outside of Google. They are for some people, who choose to adopt them as their own, but you really shouldn't point to them as the end-all-be-all argument-stopper. Google has a lot of its own idiosyncrasies. Their C# style guide puts braces on the same line, for example, because Google primarily uses Java internally.

OP used the page <title>/og:title

Was not intended as snarky. One who is still learning should not pose as an authority.

Maybe gain some programming experience before writing articles about what's considered idiomatic or good practice?

https://news.ycombinator.com/submitted?id=snapshrinkk https://news.ycombinator.com/submitted?id=Snapshrink

https://news.ycombinator.com/submitted?id=snapshrinkk https://news.ycombinator.com/submitted?id=Snapshrink

Remote playback? Isn't that literally the whole point of Plex?

Says Nix is the answer but no examples of using Nix for CI... Is this just an ad for your SaaS?

> which sounds more like a slang description of the person

Pretty sure that's the intent, since it's a play on the name "vtuber", which is used as you describe.

Avatar is used synonymously with profile picture. If I said I had an "avatar," the last thing you would think of is a fully rigged and mocapped 2d/3d avatar. Notice how I had to qualify the word avatar there ("animated avatar" is hardly specific enough), hence why it has its own name: vtuber.

Get over it. You keep complaining about the word vtuber in this thread, but it's an extremely common and popular term with a whole damn industry behind it.

AI is not intelligent. AI is not intelligent. AI is not intelligent.

You need actual humans testing your app. Only humans think like humans. LLMs don't think.

Think about what the point of user testing is. Who are your users? AI?

The MIT license is pretty simple:

> The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.

I would just put it in a comment at the top of turndown.js. That's enough to fulfil the license requirements. I know a lot of people don't care, but good to do it right. Definitely don't replace your license with theirs though :)

That looks like a really old version of turndown.js, pre-ES6. The link to turndown in the readme is the old one, too. How long ago did you make this? Anyway, you need to put turndown's license in the js file or someplace if you're not going to use npm for some reason.

Adding onto this - it might help to start by looking at how well-established tools like Buckets help to manage subscriptions. Your typical user probably has other monthly fees besides Netflix and the like, so they would more likely look for a tool that can help them track all of their monthly spending. There is definitely room in the budgeting tool market, and people (myself included!) regularly pay for those tools.

I'd suggest trying them for a few months and see how you can improve on the idea with a focus on subscriptions, if you want to turn this idea into a viable product that people will pay for.

(I hope my comments come across as the constructive criticism they were meant to be.)

So, it's a budgeting tool that doesn't do 99% of what a budgeting tool does? How is this even a SaaS? Does it pay my subscriptions for me? I could do this with a spreadsheet. Or a post-it note. If I need to cancel my subscription before it renews, I'll just set a calendar reminder.

Honestly, this feels like you were learning how to build a CRUD -- and if that's the case, nice job! It looks really good. But just taking that and slapping a pricing page on it is a bit...

If you want to create a subscription service, you need to first provide enough value to justify its cost. Although there's a certain irony in demanding a subscription fee from people worried about their subscription fees.

Btw, are those actually real users' profile pictures under "trusted by many users"? You can reverse image search them, you know; the first one is a guy named Marc. I hope you got permission to use his picture to endorse your product. If it's supposed to be a placeholder, use fake pictures.

Devil's advocate: They could make the github CLI capable of doing all of those things (if it's not already), and then the only thing the container needs is a token.