Hacker Newsnew | past | comments | ask | show | jobs | submit | mathrawka's commentslogin

You are funny. Anthropic refuses to issue refunds, even when they break things.

I had an API token set via an env var on my shell, and claude code changed to read that env var. I had a $10 limit set on it, so found out it was using the API, instead of my subscription, when it stopped working.

I filed a ticket and they refused to refund me, even though it was a breaking change with claude code.


Anthropic just reduced the price of the team plan and refunded us on the prior invoice.

YMMV


So they have no durable principles for deciding who or what to refund… doesnt that make them look even worse…?

Or they do, and two sentences from two different experiences don't tell a full story?

Okay “they do” based on what more compelling evidence?

Its not like the credibility of the two prior HN users are literally zero…


I am saying there is no evidence either way: they had contrasting experiences and one GP established this means that company has no standardized policies. Maybe they do, maybe they don't — I don't think we can definitively conclude anything.

So if you acknowledge the prior claims have more than literally zero credibility… then what’s the issue?

That I dont equally weigh them with all possible yet-to-be claimed things?


Yes, it has gotten worse as time progressed. Some k8s services started to fail, which is how I noticed something was wrong. Then the k8s control plane was up and down. Then k8s control plan completely done. Now I can't even connect to any of my non k8s servers over ssh.


Looks like the docs have a typo:

    Recommended context: 65,536 tokens (can be increased)
That should be recommended token output, as shown in the official docs as:

    Adequate Output Length: We recommend using an output length of 65,536 tokens for most queries, which is adequate for instruct models.


Oh thanks - so the output can be any length you like - I'm actually also making 1 million context length GGUFs as well! https://huggingface.co/unsloth/Qwen3-Coder-480B-A35B-Instruc...


Gotta drive traffic to that blog of AI generated posts somehow


https://pypi.org/project/llm/

I use Open WebUI for when I want a website with some more features than a terminal provides.


I really have no idea what I'm looking at...


It's not terribly obvious while skimming the page but there are some videos linked via a few of those images ("Basic SQL ▶" etc.) and the author has also uploaded loads of narrated demos over the recent past, prior to releasing the project: https://x.com/ryrobes

If I were to attempt a summary myself... RVBBIT is a SQL-powered relational programming canvas with a visual Clojure REPL and full-stack UI framework.


Project guy here - yes, succinct messaging is an issue I'm working on - there is literally so much there. I've been trying to frame it like an Iceberg - on the above the water level its a nice drag and drop "data canvas" that supports SQL and Clojure nREPL connections (meaning it could be any REPL) that generates code to be tweaked and modded for what you need it for (with some very customizable composition).

But underneath the water line it's almost like a pub/sub reactive value engine - since everything is essentially a named parameter that can be subbed to by the client (by referencing it) and the client will push updates to any subbed clients. So, run a flow - each step in the flow is it's own subbable value - run a solver (arbitrary function) - its a sub value - signals (boolean triggers) also a sub... not to mention other client's UI values... also a subbable value.

It's a deep rabbit hole (pun intended?) of reactive / cascading effects - which is a great fertile ground for interactive dash creation. IMHO data tools want to be feedback loop factories.


The dude quit his job to work on it!


This "problem" had been bugging me for a decade. The schism between "Proprietary BI Tools" (quick and inflexible) and just writing things from scratch (slow and wide open). You customers don't know or care what your tools can and can't do - and god forbid your org spent millions on Tableau (or whatever) already - your bed is basically made...

Why can't we have a bit of both - and put some direct-manipulation, Bret Victor-y Twists on it?

I just want to build "nice things" for my users that answer their questions, have some re-usability, lots of interactivity, and be able to check it into source control...

Ended up in this beautiful mutation of Tableau meets Hypercard meets Powerbuilder, etc (I think there is even a little SmallTalk and Lotus Notes in there). It's funny how the ultimate evolution of a dash tool is essentially an internal app builder...

Anyways - so I jumped off the ledge and dedicated an entire year to this - alpha is a bit rough around the edges, but it's def something interesting...

Don't let your dreams be memes - or something like that. :)


Hey man I'm not judging you if you can make money doing this!


Didn't take it that way at all. All good. :)


Sorry to be blunt, but HN is not for proofreading/vetting/reviewing your article. You should be doing that before you post it, if you are considering it to be high quality.


Mirabilis contacted you?

I worked on some ICQ software in high school, and was the maintainer of large user base ICQ clone. I still own the domain: licq.org

Never got a single job offer, guess they saw my code and went running away ;)


Thanks for licq. This was my favourite client.


Likewise, I used it for a while as well. I also used a console client named micq, good times.


Thank you for licq!


Actually Japanese quote from the Project Manager, Sakai-san, is:

> いろんな意味で新しい扉が開き、今後、これまでできなかったようなミッションができるようになるのでは。そこが一番の意義ではないか

I'm far from a translator... but it is not "We opened a door" but "A door is open to do missions that were not possible before". And he says that might be the most important takeaway from this project.

https://news.yahoo.co.jp/articles/08a8bfdd7a41486140a97a8570...


> I'm far from a translator... but it is not "We opened a door" but "A door is open to do missions that were not possible before".

You're being overly literal. Omitting the subject is normal in Japanese; translating that to a passive-voiced sentence in English is usually misleading.


Yes, but... 新しい扉が開き vs 新しい扉を開き


Again, not a distinction that carries the same nuance that it would in English. The whole animate/inanimate distinction just isn't there in the same way in non-indo-european languages.


新しい扉が開き

が makes the subject the door, not JAXA. Plus it would be very uncharacteristic of a Japanese person in that position making such a bold claim. It goes against Japanese society norms.


>Opening a new door in many senses (or "in different ways" if you want to take some liberty), from now on, missions that couldn't be done until now will become possible. Isn't it the most important (thing)?

Translation for those wondering.


Hunter2 holds such a special place in our hearts, let's keep politics out of this!


Besides, we know it's Hunter2028! No way he's running this cycle.


I'm so European I didnt even make the connection to your politics


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: