sznowicki

I had some week off work and spent some time fiddling with that side-project.

I enjoy it a lot, since I find it actually useful for discovering accounts worth following and threads worth engaging.

I hope you won't find this post spammy, here's the list of changes:

Complete redesign

Previous design was stolen from kukei.eu (web dev search index) and it didn't really fit the stuff I've been adding recently.

The re-design is fresh out of oven, probably with some bugs and mistakes but well, just YOLOed it to 'main'.

It's an effect of all-weekend work, third actual iteration and I think it's much better now.

Summaries

For certain browse categories (news, programming, technology) I take newest 1000 (or 2000) posts, throw it into an LLM model and ask for a summary in a form of "topics people are talking about".

While it's not super useful, it's nice to get there once a day and see what's up in the world. It's better than most news magazine headlines "Google made huge changes in Google Home App. We know who's affected!"

Banned

Previous update made categorization of each indexed post, with a possibility that some posts land in "banned" category (fraudulent, porn, phishing).

Today I decided to exclude those posts from search results as well (I still see creepy search queries in logs, I don't want to deal with this).

That's all folks. I hope you enjoy this piece of web as much as I do. If not, downvote the hell of this post and I won't bother you again.

💜 masto.kukei.eu/browse/news

I did some experiments over the weekend and after little polishing I think I'll keep it around for some time as a trial.

tl;dr: I process all mastodon posts I get every day on masto.kukei.eu with Ollama to try to categorize them by language and topic.

I find it quite cool and even though it's not polished, UI begs for fixes and power consumption is little too high for my liking I might keep it around for some time so feel free to check it out.

One thing I missed from my iPhone that was not really possible was to pay with a phone that runs Lineageos.

That has changed now since PayPal seems to finally offer a mastercard payments with NFC. And it works just fine on LINEAGEOS.

Jusf posting it here to everyone who are still carrying their other phones due to lack of possibly to pay easily with a phone.

I've been testing Jottacloud, aiming to find a replacement for my family iCloud setup.

I didn't find quite that, but I did find a true gem with a well-designed CLI tool for niche platforms, Apple TV app and overall good vibes.

Highly recommend them, and if you want to read more about my experience so far, and how I enjoyed talking to a support first time in years, please read the blog post.

How does it work that we eat few times a day, but usually we poop only once?

Is the colon somehow programmed to buffer the waste until the time is up and then dumps it all further?

Got this thought when observing my dog who eats twice a day (morning and evening) and more or less poops twice a day (morning and evening).

How come Americans use penguins or school buses for measuring length and distance but still use kilowatt for power and not let’s say thrown stone per mile?

I’m on vacation in Türkiya and wonder: what happens when let’s say a pregnant woman goes on vacation and for whatever reason gives birth there.

How can she take the newborn back to her country? Need to prepare all the papers in the embassy or there’s some special procedure for such cases so the paper work can be done in a country she resides normally?

One thing I miss from Twitter that Mastodon has only partially is to be able to search through the network for some keywords, like breaking news, events, and so on, so I can see other people takes on that.

So I made a small project that aims to fill this gap.

I've seen some previous attempts that made many people upset, I hope this time nobody will take their forks and pitches to burn me as it's:

• calling mastodon instances public local timelines via their API
• takes only posts that are made by accounts that OPT-IN to being indexed (there's noindex flag that I honor)
• keeps the posts in database for a short time (currently one hour) to avoid abuse and also to lower costs of running it

It gives me a lot of fun and for me it's quite useful, so I guess someone might find it useful as well.

Also, it's open source, but GPL license so it stays FOSS forever.

Happy to take any feedback, also happy to add some instances that are not yet being listened to. It turns out opt-in people are not that many so adding even large instances doesn't make my server belly up.

Instances: https://github.com/Kukei-eu/kukei-masto/blob/main/server/instances.js

So, in the era of increasingly good AI powered tools and general search engines full of SEO spam, last week I started creating something little old school and against the trends.

For now It's a have-fun-and-find-out project that main aim is to provide good search results for general web development queries with a special focus on independent blog authors.

The thesis is that no SEO spam website is in the index, which will already filter out most annoying noise on Google/Bing.

Search results are grouped per type: docs, blogs and magazines (e.g. blog platforms or bigger websites).

For now it's far from being done in terms of having a full index, but in most cases it already replaces my go-to search engine when I'm looking up some stuff during work.

I'm looking forward hearing out what y'all think and if you think it makes sense overall I can only encourage you to post some links to blogs or docs that are still missing in the index. I'm more than happy to add it to the crawler.

Responds like: "nei, total shit, who would need that" also accepted but constructive critique more appreciated ;)

EDIT: everyone many thanks for all your voices and comments. I'm super grateful for all of them and happy that we have such place like Lemmy!