@postmodern I don’t think it’s well maintained, but the TTY Toolkit was trying to fill that niche for a long time. Ruby dropped meaningful support for Tk a long time ago, and it’s often easier to throw a Sinatra front end up than wrestle with ncurses or glue UI systems together.
I prefer CLI & TUI tools, but I might be a product of the #DESQview era. I don’t even know any #programmers who still prefer byobu, vim, ctags, ack, awk/sed, direnv, and guard for a basic #RubyLang stack anymore.
Added Ruby 3.4.7 to the ruby-versions database for ruby-install users. You can now safely upgrade to Ruby 3.4.7.
$ ruby-install -U ruby-3.4.7
https://www.ruby-lang.org/en/news/2025/10/07/ruby-3-4-7-released/
https://github.com/postmodern/ruby-install#readme
@sixhat Possibly so. I mean, Python has long held more traction in the data sciences because of SciPy, NumPy, etc. Plus, as much as I personally dislike Anaconda (uv is better) it gives people an easier on-ramp than Ruby’s gem ecosystem.
So, you have a point about people using the tools they’re used to. I just wish someone would implement a sensible Ruby SDK so JavaScript could stop underpinning LLM tools and go back to implementing bad data interchange poorly.
## Please Don't Nerf RubyLang Because #StyleGuides
Ruby has made some internal changes that remind me of the changes #RSpec made to clean up internals but uglified the code & made certain features simply stop working as expected.
Removing `foo rescue nil` is an exmaple. Some edge cases make this antipattern useful; replacing a one-liner with a complex method/chain is annoying. Mindful chainsaw-juggling gave expert #Rubyists superpowers. Not trusting power users is red #Kryptonite for code.
#RubyLang is an extremely capable programming language. I've always wondered why 99% of the code for #LLM and #AI / #ML is done in #PythonLang and (for no obvious good reason) #JavaScript, #TypeScript, or #ECMAscript.
The on-ramp for Python may be easier, but it's overall less consistent and IMHO more obscure and less capable for introspection, DSLs, and for metaprogramming, so why is its ecosystem bigger for AI/ML? I wish someone would explain that in a non hand-wavey way.
The new #IBM_Granite 4.0 #Micro_AI model is now available for #beta_testing. First impressions: it's decent for its intended use case, but unsuitable for novices because it requires pre- and post-processing to avoid silly typo-induced hallucinations about imaginary products like "IBM Branite" even at Q8_0. Here's an easy fix for power users:
postmodern
Todd A. Jacobs | Pragmatic Cybersecurity---
system_prompt:
coherence:
preprocess: [autocorrect_input, fix_spelling]
postprocess: check_coherence
#ProTip: If your web form tells me legitimate domain names are “invalid” because you vibe code, laid off your experienced developers, or don’t follow the RFCs or IANA standards then your company deserves public ridicule & your your stock price deserves a minimum discount of 95%.
These companies should replace their expensive senior executives with cheap AI agents immediately. When building badly designed systems, at least automate the incompetence. Remember, bad hiring always starts at the top!
@gregly I largely agree, but *will* say that uv and especially uvx are pretty awesome. It’s kind of like Bundler but more dynamic and less manual, and uvx works a lot like Bundler without having to define a Gemfile.
That said, there are lots of apps and dependencies I can’t get working with uv, venv, conda, pipx, or any of the other Python dependency resolvers. Reach does something cool, but plays badly with something else. I think Ruby and RubyGems are simply more consistent.
We’re proud to say we called this one almost two years ago. Being a think tank doesn’t make us infallible, but we spend a lot of time thinking about the future of emerging tech. Legal battles over copyright were foreseeable and inevitable, and are certainly overdue.
The ultimate legal outcome is still murky, though. Geopolitics and monied interests on all sides will ensure that historical and legal precedents will matter less than they should, but it’s the unintended side effects on things like software licensing (deeply rooted in copyright law) that will probably have the most social and economic impact over the next decade. Expect a stronger focus on print and visual media rights this time around though.
The copyright fights are important, but a much bigger set of legal and social challenges are still to come. No matter how this plays out, it’s only the beginning.
@torgo If you have a NAS system running, setting up Time Machine on it is easy. The problem is the complexity and cost of higher-end NAS.
If you have the time, a half-petabyte TrueNAS system can be built from commodity parts for under $1k. However, you’ll trade speed for it, just like with LTO & LTFS.
DAS with RAID-6, RAID-10, or a mirrored RAID-5 is generally much faster, but you’ll pay more for hardware, disks, and proprietary software. Everything’s a trade-off.
Theia Institute: Non-Profit Think Tank