"Be happy for a fleeting moment. That moment is your life."
Remote
burakemir
@burakemir@discuss.systems
mastodon 4.7.3Interested in many things PL from theory to implementation, also logic, category theory, also distributed systems, more recently Rust.
Check out Mangle Datalog, a typed, datalog-based logic programming language and deductive database: https://codeberg.org/TauCeti/mangle-go http://codeberg.org/TauCeti/mangle-rs
#datalog #CategoryTheory #logic #types #systems #QueryLanguage #DistributedSystems
236 Followers
77 Following
11 Posts
Joined August 07, 2023
GitHub:
codeberg.org/burakemir
Homepage:
You may have heard about the 4G model your Chrome downloaded. Here's how to get rid of it - via LinkedIn, Himanshu Sharma posts:
---
1) In Chrome, click the address bar and type: chrome://flags then press Enter.
2) Turn off the key AI flags by disabling the following:
>> optimization guide on device.
>> Prompt API for Gemini Nano.
>> Prompt API for Gemini Nano with Multimodal Input.
>> Writer API for Gemini Nano.
>> Rewriter API for Gemini Nano.
>> Proofreader API for Gemini Nano.
>> Summarizer API for Gemini Nano.
3) Relaunch Chrome.
4) Delete the whole OptGuideOnDeviceModel folder.
---
That's ~/Library/"Application Support"/Google/Chrome/OptGuideOnDeviceModel/ on my machine. Good riddance.
Good description of the Read-Copy-Update (RCU) pattern https://www.infoq.com/articles/read-copy-update/
This morning I saw @djg@mastodon.world in my news feed, an interview on the impact of AI software engineering. I had met Dan maybe 20 years ago when he visited EPFL and gave a talk about Cyclone, one of the early languages aiming to provide memory safety for low level, address-exposed programming. It was good to read some balanced, grounded views, and I share the optimism: most skills remain relevant and we haven't even gotten close to what is possible with software. The interview: https://www.businessinsider.com/cs-professor-explains-changes-software-engineering-ai-2026-4
Replying to
@maxsnew in a CS context, it might be better to spend the effort on developing CS examples.
Replying to
@zanzi What is clearly problematic is the loss of skill and the vicious loop where everything original will be "captured" and assimilated in the next model without any attribution. It is a dynamic that is similar to open source software, the value generated by human creativity and effort gets monetized by few. That might be the biggest issue. When it comes to science and education, the big opportunity is in cheap and effective dissemination. It is maybe less the novel results but that a larger number of people can more easily understand and apply already existing results where we should look for positive outcomes; if we can define the standards of quality and verification for the things we care about.
Replying to
@koronkebitch great post, if slightly pessimistic. Towards the end of my thesis, I came toward a similar realization that the scientific process comes down to interactions of people. Many things require a look beyond the homo oeconomicus perspective and real engineering requires a concern for safety and sustainability, not only efficiency. Yet, there is also a place for looking at economic factors because solving problems for humanity does require looking at cost.
"Human beings, viewed as behaving systems, are quite simple. The apparent complexity of our behavior over time is largely a reflection of the complexity of the environment in which we find ourselves." Herbert A. Simon, The Sciences of the Artificial
#Mangle #Datalog news: mangle-rs has python bindings now. Get them on https://pypi.org/project/mangle-py/
Let me know if there are any problems or you have suggestions for the API.
As a reminder: Mangle is a datalog based language, with a go and a Rust implementation. The Rust one can also conveniently used in the browser, via WASM.
The time to use datalog from your code is *now*, no excuses 😉