Hacker Newsnew | past | comments | ask | show | jobs | submitlogin
Show HN: Every open source tool from the "What's HN working on" thread (github.com/getomni-ai)
169 points by themanmaran on Aug 28, 2024 | hide | past | favorite | 34 comments


Since this is a CSV on GitHub: https://github.com/getomni-ai/datasets/blob/main/hn_projects...

You can open it in Datasette Lite (WebAssembly in your browser) to run SQLite SQL against it like this:

https://lite.datasette.io/?csv=https://github.com/getomni-ai...

Here's a SQL query showing just the open source projects focused on AI, for example: https://lite.datasette.io/?csv=https://github.com/getomni-ai...


how is this opening it? i am getting cors errors on my tool, are you doing some server side processing?


answering my own question. github raw content does not have cors problem. You can construct the raw content link from the normal github content link to download using fetch.


THIS IS FN DOPE.

---

I need more examples of datasette!

Thanks for showing this.


Turns out I've shared a whole bunch of examples like this in comments here in the past! https://hn.algolia.com/?dateRange=all&page=0&prefix=false&qu...


https://github.com/SoMaCoSF/CyberScraper-2077

Need to figure out how to connect Datasette with CyberScraper.

---

Heh - I forgot to look at your username. :-) long time fan.

BTW I recall when you guys announced Lanyrd on YC.


Hey I am the developer behined cyberscraper-2077, I will look into your feature request :)


Cool stuff!

You might consider an alternative to screenshots for tables. There are a few problems with this strategy. Most significantly, it's inaccessible to the visually impaired - they're hearing your description of the data without getting to read the data itself, which must be frustrating. Problems like this can feel hypothetical and abstract, but HN is a large enough audience that there are probably real people having this experience.

It also can't be copy pasted, it can't be plotted, it doesn't respect the settings of my browser like font size, etc. When I see some of these breakdowns, my first thought is, "hmm, I'd like to copy paste this into a spreadsheet and make a bar chart." (Indeed, that's when I noticed it was a screenshot.)

You might consider copy pasting from DataGrip into your editor and formatting it to be a markdown list or something. It probably would take about the same amount of time as taking a screenshot and putting it into your markdown.

Thanks again for sharing, this was an interesting breakdown.


You know I agonized over this for a bit before giving up and adding the screenshots. Originally I was trying to create some bar charts as well but decided to just post and edit later.

Planning on a better writeup later and will incorporate this feedback.


For sure. Done is better than perfect and iteration is a great strategy. I wouldn't have made this criticism if I'd realized you agonized over it.


Recent and related:

Ask HN: What are you working on (August 2024)? - https://news.ycombinator.com/item?id=41342017 - Aug 2024 (1365 comments)


Also, from today:

Ask HN: Who's building an AI-free product? - https://news.ycombinator.com/item?id=41375100


Pretty cool, nice to have an easy to read list.

One issue I noticed with my project [1] is that the description is for the wrong thing. This is probably partially my fault, since I wrote about two projects in one comment [2]. The music production app is closed source, but the UI library I maintain for it is open source. Seems like that might have thrown off your AI tool.

[1] https://github.com/Pulselyre/UpbeatUI

[2] https://news.ycombinator.com/item?id=41342977


I noticed this was the case for the project laudspeaker as well. The description says

"Techno Thriller Panopticon Explores Encryption and Espionage"

but the project is for customer engagement and onboarding.


Oh yea the titles are a little off, since they're pulling off of the original comment and not the repo itself. The Laudspeaker description came from:

"Apart from working on my start up, and open source project https://github.com/laudspeaker/laudspeaker (open source firebase cloud messaging) I've been writing a short sci fi story, sort of like a techno thriller inspired by michael crichton, called Panopticon."

https://news.ycombinator.com/item?id=41348190


good catch - yes I added two projects in one comment!

laudspeaker (open source firebase cloud messaging/customer engagment) - https://github.com/laudspeaker/laudspeaker

panopticon (short story on encryption and espionage) - https://docs.google.com/document/d/1VRI4X5fCUpwurUDvKmvzJpT7...


This is awesome, thanks for doing this! Did you use Omni for the sentiment analysis?

Now, for the analysis of the Who’s Hiring posts.


As submitted 2 days ago with an extensive sales pitch but may be useful. A self second chance queue submission, perhaps!

https://news.ycombinator.com/item?id=41351839


Some Show HNs are linked to sites/blogs with open source links. Next iteration of this can include those.


Oh yea. I wanted to do some recursive web scraping to pull more info.

hn_comment => demo_link => github as well as hn_comment => github => demo_link

One of these weekends I'll pull the last year of this thread and get a full dataset.


This is amazing! I've been meaning to do something similar for all the Show HN threads, granted it's a much bigger set, but I haven't had the chance to.


I don't think it's a good idea to give a DB connection string to everyone, even if it's read only


Funny enough, looking at the db logs, no one queried it. I think simonw made it so much easier with Datasette that no one bothered.

But I did spin up a single use free tier supabase db for this usecase. Spent a lot of time worrying some hacker would find an exploit.


Nice, this is cool :) Thanks for sharing


FYI some of the GitHub links return 404.


This is awesome! Is this using Omni?

Would love to see this for more topics, like products people are working on


Yup! And this was pretty quick to pull together. I'll definitely be publishing some more of these type datasets.


why is the description for https://github.com/madprops/goldie is "Firefox Vertical Tabs Powerful Python Chat Client Nim Text Finder" ?


It got those from this comment and transposed the descriptions: https://news.ycombinator.com/item?id=41347438


This is my favorite part about software


Great idea! Thanks for doing this!


Not sure how your code is identifying "open source", but you missed including FreeBSD on the list.


It got tagged as open source, you can see it in the csv [1]. But the original comment didn't have a link and table I shared was filtered to just posts with links to the repo.

[1] https://github.com/getomni-ai/datasets/blob/main/hn_projects...

[2] https://news.ycombinator.com/item?id=41342592


Hey! Cool!




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

Search: