tom_usher

I definitely appreciate all the work that has gone in to LiteLLM but it doesn't take much browsing through the 7000+ line `utils.py` to see where using it could become problematic (https://github.com/BerriAI/litellm/blob/main/litellm/utils.p...)

Seems to be a change in Cloudflare's managed WAF ruleset - any site using that will have URLs containing 'camel' blocked due to the 'Apache Camel - Remote Code Execution - CVE:CVE-2025-29891' (a9ec9cf625ff42769298671d1bbcd247) rule.

That rule can be overridden if you're having this issue on your own site.

Thanks, this is an important nuance. Recommendation algorithms are absolutely useful, and if you're so inclined you can absolutely make them work for you, but this is about making educated, conscious decisions about what you click next in your 'Related videos' section.

Algorithmic feeds don't give us that opportunity - they're designed to require minimal effort and to keep the dopamine coming without any conscious decisions.

Great point. I'm personally trying linkblogging and following other link blogs inspired by Simon Willison [1].

The more people that do this the more we can start rebuilding networks of people we trust and still retain control over the diversity of our sources.

1: https://simonwillison.net/2024/Dec/22/link-blog/

I was definitely influenced to write this by Alec's excellent video which I recommend everyone watch.

I'd hoped it would be a way to share my own opinions on it, summarise my own personal concerns, as well as adding my own recommendations - but totally appreciate if you feel it is derivative, and I appreciate the call out. As a big Technology Connections fan I certainly don't intend to steal his work.

It's also intended as something you can link to your friends and family that might be a little more digestible than a 30 minute video.

This looks very useful and far less hacky than my solution using OBS' Browser source and Bodypix/Tensorflow.js (https://usher.dev/posts/make-your-webcam-look-slightly-more-...).

Can't get it to work at the moment - but hopefully with a bit more experimenting, this will replace my current options.

Hadn't realised until reading this submission that Zoom has added blur support on Linux recently - on trying it out it does seem quite poor though - a very sharp edge and quite slow to update.

Excellent! It's great when it tries to generate code: https://stackroboflow.com/#!/question/8138 (the last line here made me laugh)

This is great - any others you can recommend?

Fair enough! I'm intrigued to know your opinion of the game from what you've played of it. Do you enjoy the synchronization/counting aspect?

I think that's a little reductive - there is more to it in the same way that there's more to poker than hoping you get dealt a good hand (I'm sure that's opening a can of worms).

It's not a game designed purely for expert bluffers. Everyone I've played The Mind with has their own tells and quirks - even when they know the rules forbid it - they'll cringe at their hand, tentatively hold a card forward, or lower it to the table. It's just fun to pick up on these and have a laugh about them.

I'm sure there are people that will rules-lawyer the fun out of it, but with the right crowd this is one of the games I've been pulling out most frequently recently, and haven't heard a bad thing.

I've never been able to find the one note-taking system to rule them all, but that seems to be because the contexts in which I need to write/access notes are quite different.

At the moment I've settled on a three-tier system:

NVAlt - as a scratchpad, clipboard buffer and other transient notes.

Moo.do - for structured notes, Todo lists, reminders and agenda/scheduling

Dokuwiki - for long term knowledge base stuff and for documenting personal and home projects. Love this because my family can easily read and use it too.

I've tried loads of tools but I've stuck with this set longer than any other so far.

I'm a big Moo.do[0] fan - I use it for notes, my todo list, and agenda, so at the moment I've got it set up as my new tab with some custom CSS to hide the stuff I don't need to see in that context.

This seems to distill the concept down very nicely - quicker to load and straight to the point. Sadly I'd miss the structure of Moo.do but I can definitely see a world in which I'd use it - nice work.

[0] https://www.moo.do

I've been seeing an unusual amount of NTP requests in my PiHole logs but never got round to figuring out the cause - nice to have an explanation.

Hope this is fixed in a firmware update, my repeater is quite a nice device otherwise.

For what it's worth, I use Insomnia every day with some pretty large requests and I've never had an issue with it.

With this, VSCode, Slack & Boostnote - a good portion of my daily work is done in Electron apps and I can't say I'm itching to switch to native alternatives.

Definitely appreciate how people have issues with it, but it doesn't seem like it's going away and hopefully will only get better.

Ambitious is definitely the right word for this game. I've put about 7 hours in to it so far and can definitely say I've not played anything quite like it. It's huge, full of unique art and writing, and has managed to surprise me quite a few times.

I can see it getting repetitive, particularly as there's not a huge amount of storytelling - and from what I've heard, some people aren't enjoying the need to replay hours of content when you die - but so far, I'm impressed.

The second printing (+ expansion) Kickstarter is live at the moment (https://www.kickstarter.com/projects/1926712971/the-7th-cont...) which seems like it's adding a lot more content. I'm in, but whether I'll still be interested in playing more when it ships in October 2018 is another question...

I've not come across anything that I would call 'epic' from a minimal set of rules, but some of my favourites that you can throw in your bag, teach in 2 minutes and play in 15:

Welcome to the Dungeon/Dungeon of Mandom - https://boardgamegeek.com/boardgame/150312/welcome-dungeon

Love Letter - https://boardgamegeek.com/boardgame/129622/love-letter

Hanabi - https://boardgamegeek.com/boardgame/98778/hanabi

No Thanks - https://boardgamegeek.com/boardgame/12942/no-thanks

Archaeology - https://boardgamegeek.com/boardgame/191300/archaeology-new-e...

Linko/Abluxxen - https://boardgamegeek.com/boardgame/153065/linko

This looks nice and simple!

I've been enjoying Lost (https://github.com/peterramsing/lost) with PostCSS for grids recently - it has Flexbox support which works well and keeps grid class names out of my markup.

Great work, I've been enjoying using this exact stack recently - this has definitely taught me a few things.

The main thing I'm still trying to get my head round with React/Redux/Router is where in the lifecycle to do data fetching - I noticed you're doing it in componentWillMount, when the React docs suggest componentDidMount, and I've seen others do it in various other ways with no clear indication what the 'best practice' approach is.

Sorry for the nausea - it's only after posting that I realised I could have been less antagonising. It's good to hear your opinions though.

Really interesting to hear your experiences - thanks for sharing.

I wouldn't call it zealotry, but probably resistance to change.

As with anything new and interesting, it's easy to insert a fictional future scenario that takes away what makes our personal lives rich and enjoyable.

I love to cook but it's all a bit sci-fi and uncomfortable to think that we'll all have our government mandated NutriSludge dispensed twice a day so we can all stop worrying about cooking and work harder.

Thanks for the answer (and to everyone else that replied) - it's always good to hear the other side of the debate.

I could never imagine myself in a situation where I'd consider it so important to be constantly busy that I wouldn't be able to find the time to throw a few vegetables in a pan - but I understand different people have different priorities (and demanding jobs).

It just seems to me like this is an odd way to solve the problem, maybe because I'm extrapolating a bit and seeing it as trying to replace one of the things that makes us human; the ability to prepare and cook food.

I'd be interested to see if any research comes up as to the long term psychological effects of this sort of product. Do people who replace the varied textures and tastes of different foods (if only partially) with a liquid supplement notice any adverse effects or cravings?

Out of interest, is Soylent more interesting than offerings from companies like Graze/Naturebox?

Can someone explain to me the appeal of this product? Why would anyone choose to drink a nondescript sludge over the plentiful ways we can get good, enjoyable & healthy food?

I can certainly appreciate the science behind it - this would be a great boon to helping feed those that are less fortunate, but from what I can see that doesn't appear to be the aim of the company.

I've been following Ansible from the start, Michael has always seemed like an excellent lead for the project - he's had a strong opinion and vision for the project - Ansible has turned out really well as a result.

More recently, Ansible Inc's faceless approach to marketing has me a little frustrated, but this isn't the time for complaining.

Thanks for all your hard work on Ansible Michael! I hope the team you've built can continue to pursue your vision for the project, and best of luck wherever you end up next!

This looks great, nice work.

For the dev/prod parity use case, where do you usually draw the line between Packer and Vagrant? I'm currently building boxes with Veewee with all dependencies pre-installed and distributing this to the team. Vagrant could easily deal with this, installing dependencies when the VM is launched using provisioners.

Does anyone have any preferences/reasons to go one way or another?

I've enjoyed using Peewee[1] as a lightweight ORM for Python - mainly for scripting data manipulation on legacy databases but it does the job nicely. How does Pony ORM compare?

[1] https://github.com/coleifer/peewee

Slate is my preferred way of managing windows on OSX.

https://github.com/jigish/slate

Slate also recently started supporting configs written in JavaScript (https://github.com/jigish/slate/wiki/JavaScript-Configs) which is very powerful.

It's perhaps interesting to note that at the size I run my browser (half my screen horizontally), the 'Sign Up' button is cut off - maybe responsive is slower to implement, but at least it makes you think about issues like that before launching your new site.

Nice, but why tell me to 'Avoid using British English.'? Maybe a one-off misuse could be caught, but if you see multiple instances of British English in a text block, maybe you can assume the writer intends it to be that way?