HN user

msm_

1,100 karma

tailcall.net @MsmCode github.com/msm-code @msm@infosec.exchange

Posts6
Comments389
View on HN

Literally the main point of this blog post:

The project’s binary distributions didn’t include the required MIT license notice for the llama.cpp code they were shipping. This isn’t a matter of open-source etiquette, the MIT license has exactly one major requirement: include the copyright notice. Ollama didn’t.

Transcribe.cpp 3 days ago

It sounds like the only way this would make sense is if such model knew the range of sounds it expects to "hear". There's a lot of possible sounds that IPA knows about, but world languages only use a fraction of them at once. Think English dark and light "l" (ball/light) or aspirated "p" (pin/spin) - some languages contrast them, while in english the difference is not meaningful.

Or maybe linguists are actually interested in having maximally faithful IPA representation and manually normalizing it? You are clearly way more knowledgeable about that topic than I, so I'm curious what you think.

GP said:

you shed a layer of security by deciding to make your address public, non of which would benefit the whole point

It's possible to host something with no intention of making it internet-public. I also have services like that, that I only use myself or with friends. GP argument is that they don't want to share the onion link to their website, because (bluntly) we are not invited. Onion domains are actually relatively private (i mean unguessable - unlike clearnet domains), so it's possible to host private websites without any additional authorization.

Having said that, onion links carry a implicit baggage, so while I think they're great for sharing things with (technical) friends or a private VPN, they're probably not the best way to host services intended for public.

GitRoot 4 days ago

In GitRoot everyting is stored in git, not in a database, not in a hidden blob in your git tree. Everything is stored in plain files aside your code.

(...) If you add a user, by editing .gitroot/users.yml or by merging a branch where a user has added itself, this user will be able to push to the default branch. All GitRoot features are articulated around this concept (...)

Regressive JPEGs 5 days ago

I personally don't remember any visual progressive image buildup in like decades, so it's not doing anything valuable at all.

Maybe you just don't notice? It can be pretty invisible sometimes. I sometimes notice that image soon after page load an image is slightly blurry, and then another pass "sharpens" it. Yeah it's not like in the "old times" when the first progressive level was almost unreadable, but there's still value in sending a lower resolution version of image in 30% of the total file-size, basically for free

Just as a datapoint - I really like learning with Anki. If there was another method that would let me learn at 200% speed, I would pick Anki.

This is moot anyway, because most other learning methods require longer focus and are not possible to do while on the go, during short commute in a tram, or 3 minute wait in a queue, and much more.

I exclusively read fiction, mostly fantasy books (I used to also read programming books, but all my learning is online now). I enjoy them. I don't feel the need to "improve".

Because of my job (and hobbies), I spend a lot of time thinking and straining my brain. After 14 hours of working I don't feel the urge to read Nietzsche or Plato. I admire people who do, but for me it's just a way to relax (and maybe escape from daily chores). Am I doing it bad?

Of course, just like it instilled a desire to consume a lot of calorie-dense food. "Desire to protect the children" in this case is a knee-jerk reaction, or a thought terminating cliche.

For example, how many children will this actually protect? How many children will this harm? How many adults will be harmed by inevitable side-effects that arise? Those questions are not discussed or even considered.

Regulators are responding to a real problem: EU-funded research estimates driver distraction plays a role in 5% to 25% of car crashes

Article 6(3) of the GSR states that the system should be designed in such a way that it does not continuously record or retain data other than what is necessary for its purpose

I get that there are problems, but it doesn't sound that bad to me? Car drivers kill tens of thousands of people every year in Europe. If we can improve this 25% (more realistically, 10%) it's a huge step forward.

Linux capabilities have many problems (they are too coarse-grained and too many capabilities are root-equivalent). But anyway this is an overkill in this case probably. In may distributions access to /dev/kvm is guarded by membership in the kvm group - no need for new capability, just regular old filesystem permissions.

There are A LOT of cheaper ways to get some writing samples on the internet. This is just someone having fun, and creating an interesting challenge for others to have fun with.

I don't know, it looks pretty decentralised to me?

The purpose of this list is to allow IPFS node operators (e.g. someone running a public IPFS gateway) to opt into not hosting previously flagged content.

IPFS node operators, who are supposedly interested in hosting malicious content (and i2p-hosted phishings are a real problem) can OPT INTO using this list.

In this case, I don't see how that's any problem for piracy - people can just use one of the bad/unfiltered nodes.

Average monthly salary in Nigeria is $220.42. Median monthly income is $65.02. Minimum wage is $0.26/hour.

I know this kind of money sounds banal for us now, but even I, when I was a student (especially in high school), tried to avoid spending any money on my hobby (programming).

Google has over 100,000 employees, which means statistically speaking a few of them have committed or will commit homicide

Nitpick, but the set of people who work at Google is highly non-random. I am pretty sure Google employees are much less likely to confirm homicide - they are older (homicide rates after 24 drop sharply), more educated (strongly inversely correlated with homicide), richer (same), etc.

there was the open source office for approving open source code (which amounted to a rubber stamp in my experience; they mainly checked for boilerplate issues)

In 2018 I applied for a permission to work on an open-source project, in my own spare time, and I was denied. The problem was the license (AGPL is like poison to corporations), but it's my after-hours side project anyway so I still don't get why it mattered.

No the system is not perfect, but it is still one of the best in the country, except for NYC and maybe Boston

I mean, there are a lot of poorer countries (especially in europe) that manage to solve this in a much better way, so this kind of proves OP point that raw purchasing power is not equivalent to the standard of living.

uBlock Origin Lite is still working and will continue working (since this is a MV3 issue). It's worse than the original version, but it works.

(I use Firefox, but I've read your post before the linked post and I thought ublock Origin is the one being killed)

Are people on HN still typing out functions by hand one character at a time?

Well I use tab completion, of course. And I copy-paste snippets from LLM more often than from SO now. But otherwise not much has changed in my career in the last 5 years. Is this different for you?

I'm not fundamentally opposed to code generation, and I use LLMs for some taks, but I don't see myself vibecoding whole pages of production code. I vibecoded a throwaway note-taking app for myself though.

Gitlab is a security nightmare. Self-hosting Gitlab is pain and a lot of work. Or course you can neglect security if you don't expose it to the internet, but it's not exactly a dream.

I'm all for self-hosting btw, we do it at my company. But it's not as easy as you make it sound.