Clojurians Log v2
off-topic
Has anyone ever leveraged semantic (git) diffs to view the changes to a specific function/form over-time?
Makes so much sense its a clojure/datomic project! I figured it very much touches the code=data idea
This indexes repos in Elasticsearch for semantic/vector search. I don't think it does anything with git or change over time though: https://github.com/elastic/semantic-code-search-indexer
Yesterday an “official expert” from Gartner said to me that if we want to stay relevant in 5 years we should start building Agentic AI orchestration. Then, silly me, I engaged politely with one random ex-Clojure guy on social media who was saying that Clojure is a dead language and he told me “stop wasting my mentions”. Now I am not sure whether to start building Agentic orchestration in Python, or stop talking to people.
inconceivable that llms are not even here for that long and people are already calling themselves experts
I see most AI orchestration fandom coming from people that have no experience cleaning up AI slop ...
If you want to be relevant, know how things work, know how to read docs, be useful
For at least the first 20 year of Python, few regarded it as a viable language choice outside of Academia (and some Linux tools). A similar story for JavaScript. Clojure just turned 18 this year. Its having a data science conference this year along with many other conferences across the world. I nearly did a Phd on 'Agentic AI orchestration' (we just called it software agents) back in 1997. There was much discussion at that time, especially regarding all the things that could go wrong and the risks it could expose individuals and companies to. Talking to a wider range of people may help get better view.
my take, being involved in this agentic way of developing with Python at work, is that it seems designed to overwhelm the human brain: we can (still?) not fully delegate to machine code creation, but we can now produce an avalanche of code in a few key strokes. Essentially both the prompt creator and future reviewers cannot really handle the load. Plus a false sense of confidence is created by the fact you can generate the tests themselves for you feature (which is more code to review). Also now while you wait for a review, you can produce other N avalanches of code to open a PR with. I would suggest: protect your capacity to think and reflect (that is hard!) and for sure you will be a wiser and more pleasant human being. That will always be relevant even if computers will stop working tomorrow.
welcome to the future, where every question hinges on something important already being dead, or dying soon if you don't change something drastically