fareesh

liveview feels a bit too magical

client disconnects, state desyncs, then reconnects, then liveview figures out via crdt and other things how to accurately diff the state

feels like i have to trust it too much - i'd have a lot more confidence if people were vocal about how they are running it at scale, battle-tested etc. etc.

shopify is straight up yolo running rails edge, which is a crazy endorsement

from what i understand:

- if you compile from source and deploy via adb nothing changes

- if you use a closed source binary, the identity of the owner becomes mandatory

so the issue is anonymously published closed source software?

I've read both sides of the story and it does not make sense why he would change the root password and not communicate this with RubyCentral or make any attempt to contact them about the change. That in itself suggests malice.

they were funded by a company run by a guy who read DHH's blog and cut funding unless they excommunicated him

Can't seem to connect to qwen.ai with DNSSEC enabled

> resolvectl query qwen.ai > qwen.ai: resolve call failed: DNSSEC validation failed: no-signature

And

https://dnsviz.net/d/qwen.ai/dnssec/ shows

aliyunga0019.com/DNSKEY: No response was received from the server over UDP (tried 4 times). See RFC 1035, Sec. 4.2. (8.129.152.246, UDP_-_EDNS0_512_D_KN)

i agree with all that

but the stuff we're seeing is akin to not liking the same flavor of ice cream

soft, mild, first-world problems

i use node/npm moderately

is there a runnable command to determine if the package list has a compromised version of anything?

the Android change doesn't impact your ability to plug in your own device and run your own code or someone else's code

the change impacts closed source software distributed without verification which is by definition unknown so the "want" is not possible - i.e. you can't know if you want to run it.

it's very common in india

I guess you could just try an equivalent in chatgpt or gemini or something. Paste 5 text files one after the other in some structured schema that includes metadata and ask a question. You can steer it with additional instructions like mention the filename etc etc.

short answer is that in rag systems the documents are chunked into some predefined size (you can pick a size based on your use-case) and the text is converted into vector embeddings (e.g. use the openai embed API) and stored in a vector database like chroma or pinecone or pg_vector in postgres

then your query is converted into embeddings and the top N chunks are returned via similarity search (cosine or dot product or some other method) - this has advantages over bm25 which is lexical

then you can do some processing or just hand over all the chunks as context saying "here are some documents use them to answer this question" + your query to the llm

isnt it a rails mpa with some modern layers on top?

don't they all sign a document saying it's all classified tech?

what are some automation use-cases for something like this?

feels like a cool project/toy

Is there a list of popular attack vector urls located somewhere? I want to just auto-ban anyone sniffing for .env or ../../../../ etc.

Rather not write it myself

I have a similar setup - my post goes into some more details on getting a few things working like embedded tweets etc

https://fareesh.com/post/config-to-content/

i continue to be amazed by the l33t h4x0rs who are caught because they forgot they have russian ips

The non english space in these fields is so far behind in terms of accuracy and reliability, it's crazy

in my experience people who are on the political left are very rude and dismissive of any heterodox position as some moral sin

"async_visit_link for link in links omg it works"

imo the primary benefit of react is the nice component library.

emdashes are on the rise thanks to people copying and pasting chatgpt

one of the worst culprits of this is bytedance's scraper which has a singapore ip block

amusingly some scammy companies have crowdfunded off this type of "traffic" by citing "interest in asian markets"

[edit] seems like it's federated

- Where does one get an affordable server that isn't on a blacklist somewhere?

- What happens when one of the big cloud providers arbitrarily start putting your emails in spam?

Are there solutions to this? It feels like the biggest value provided by "big email" are these two things

possibly vandalism by someone looking to be the next Vindman-esque hero in their discord group

Did you give inertia a try?

I do this too, it works well

What solutions are folks using to solve queries like "How many of these 1000 podcast transcripts have a positive view of Hillary Clinton"? Seems like you would need a way to map reduce and count? And some kind of agent assigner/router on top of it?

Prediction: The Onion will be out of business within 5 years