HN user
co-founder and co CEO www.discourse.org hobby: term-llm.com
My agent wrote a pile of very interesting articles at wasnotwas.com I have been a bit quiet there for a bit, but it covers lots of areas that are very interesting to harness builders (albeit less interesting to the general public)
I did a follow on analysis with got 5.4 and opus 4.7 https://wasnotwas.com/writing/claude-opus-4-7-s-system-promp...
For those interested, edit is a surprisingly difficult problem, it seems easy on the surface but there is both fine tuning and real world hallucinations you are fighting with. I implemented one this week in:
https://github.com/samsaffron/term-llm
It is about my 10th attempt at the problem so I am aware of a lot of the edge cases, a very interesting bit of research here is:
https://gist.github.com/SamSaffron/5ff5f900645a11ef4ed6c87f2...
Fascinating read.
Codex requires stuffing a very specific system prompt otherwise the custom endpoint will reject you
Author here, thanks heaps for the discussion, I replied to a few of the points in my blog comments:
https://discuss.samsaffron.com/t/your-vibe-coded-slop-pr-is-...
Qwen coder 32b with a JavaScript interpreter
Impressive answer for a model that can run on your own computer
https://discuss.samsaffron.com/discourse-ai/ai-bot/shared-ai...
I find it odd that is refused me so badly https://discuss.samsaffron.com/discourse-ai/ai-bot/shared-ai... my guess is that I am using a quantized model
It simply did not want to use XML tools for some reason something that even qwen coder does not struggle with: https://discuss.samsaffron.com/discourse-ai/ai-bot/shared-ai...
I have not seen any model including sonnet that is able to 1 shot a working 9x9 go board
For ref gpt-4o which is still quite bad https://discuss.samsaffron.com/discourse-ai/ai-bot/shared-ai...
The original was posted at work earlier this week, to me the original missed a bit around explaining what this tech is yes good at... https://meta.discourse.org/discourse-ai/ai-bot/shared-ai-con...
Highly Gamed === It is better if users with slow devices see a white screen for 30 seconds vs an indication that something is happening, because ... reasons?
For those looking for a rubyish approach to this see: https://github.com/discourse/mini_sql
This is something we are investigating. We would like it to interoperate with the wider ecosystem.
Link with installation instruction is at the bottom of the blog post:
https://meta.discourse.org/t/discourse-chat-plugin/230881
You can try it out meta.discourse.org if log in to meta and then join the https://meta.discourse.org/g/chat-testers group (self serve)
Regarding bundle sizes, we hear you and have some long term plans to investigate code splitting, it is a very long journey but we will get there. At the moment you get the majority of the app on first payload, making subsequent clicks very fast.
Could you include a link to the new podcastified Twitter space in the footer of https://oxide.computer/ ?
And maybe do an addendum 5 minute announcement episode on the “on the metal” podcast?
Love your excellent work, thank you all
Discourse | Remote - UTC+-3 | Full Time | Infrastructure Engineer
Discourse is a profitable, open source company, we host for thousands of public forums and internal team sites.
Our infrastructure team is responsible for taking care of all our hosting concerns including monitoring/deployments/alerting and so on. Much of our work involves amending and improving automation.
We are hosted on bare metal Linux servers and AWS.
To apply and for a full job desc see: https://www.discourse.org/jobs
Is there an app on iOS that indexes ubuweb, I noticed Junky read by Burroughs there which would be a fascinating listen, but it comes in lots of pieces and consuming it on the run seems somewhat tricky
Some experiments are underway soon, current implementation uses a cache that is tightly scoped to a specific run including stuff such as class serial numbers that can not be reused between runs safely
The terms do get a bit tricky, with forking servers like "unicorn" that do copy-on-write memory we also have the term PSS which is a far more accurate was to calculate memory impact for such servers.
The difference between the RAM your program uses to PSS to actual memory your process can reach vs the memory your program has allocated can all make your head spin. Agree it makes sense to add a section to illustrate this.
OP here, if you have any questions or tips for other similar tools you use I would love to hear!
I did see quite a few that double up, like 2 versions of url and a few others.
Just had a quick play ... on arch linux `yay nushell-git` already works!!!
One area where I got a bit stuck was around help. `man where` gave me nothing, `help where` also. I tries stuff like `ls | where type = File` and got a type error. I think it would be amazing if this thing onboarded people a bit nicer, "where needs a condition, to learn about where type "help where" ... stuff like that
Overall really enjoying the ideas here and I am absolutely going to be following this!
There are 2 parts to the cost.
1. is the VM 2. is handling filesystem mounts across operating systems
(2) can be debilitating in dev environments on Mac for example where you want to edit files outside the container but have the container have access to them. Docker have done a lot to improve this over the years but it is still very painful.
absolutely if you have a filesystem heavy operation and are accessing files inside the container in aufs instead of a local mount, you are going to pay a massive docker tax.
A very interesting and important aside is that in Linux you pay almost no tax for using Docker.
This is not the case on Windows and MacOS where there is a significant cost.
That 8 minute spec run, also takes 8 minutes in docker.
https://meta.discourse.org/t/beginners-guide-to-install-disc...
I completely understand the concern here. If every change required an 8 minute wait it would be real sad for us.
We use a tool called `rake autospec` at Discourse which is a smart test runner. I personally like the vim integration we built.
The way it works...
1. Run `bin/rake autospec` in a terminal 2. I head to the code I want to change, I change the code 3. The spec runner figures out what the right test is run and runs that. 4. While it is busy running specs at any time it can be interrupted by saving a "*.spec" file in which case the spec runner will run the spec at the cursor
This has worked wonderfully well for us.
Absolutely disabled defender.
When I started running the spec suite in WSL, defender went way up (so I gave it a break), pretty sure stuff would have been way worse with defender on.
The prevailing recommendation though from the WSL team is to leave defender on for now. So I kind of cheated by disabling it.
Unless you are looking to contribute to a project and kick off a project I would recommend sticking with the bigger players our there like nodebb/vanilla and so on.
Getting stuff right is not easy, very few forum software platforms out there for example have a bug bounty like (https://hackerone.com/discourse ), last thing you want to do is deploy an XSS hive out there to the public.
As far as I know there is no small-medium well supported Discourse alternative written in Go. There is a Slack alternative though written in Go called mattermost which I can recommend.
WSL is not technically a VM...
The news is not the it is slower, it is how much slower it is. I found the number surprising given the prevailing wisdom out there was ... sure its slower, about 5-10% slower.
The other news is that mac is slower and it does not do any virtualising.
Tracking the MRI enhancement at: https://bugs.ruby-lang.org/issues/15667
For us at Discourse we find Mattermost perfectly acceptable as a chat program.
10 second load times and slow/wrong notifications have not been a problem for us at all.
really depends on age, my 5/8 year olds love 5 minute dungeon, very simple and quite fun. hanabi is a bit too tricky for them. codenames pictures 2 players is cooperative cause you play against "the computer" and is quite fun.