my personal theory is that it's a cultural problem. same reason web apps are such memory hogs. front end web devs, as a culture, do not care about performance, but they really care about developer experience. the opposite is true of systems developers
Note that Obsidian parses all the files and uses the browser's built in database to store indexed metadata and links between note files. It's how they have a responsive search function. You could either use sqlite, ad-hoc your way with flat-files.
Of course, you could parse, walk links, and search through all files on the fly all the time. Text files at personal-notes scale should cache really well, and it might still be faster without the browser-tech overhead, but it's a toss-up
That point is when you have a human expert validate that the ai-generated code is correct. If the community as a whole has given up on doing that, it does not retroactively make all LLM-generated code not-slop, it just means slop is the norm.
And obviously from the reception you've gotten, even that has yet to occur.
personally, I find it useful as a crude heuristic for identifying software that probably won't make someone wake me up in the middle of the night because it crashed because of some ridiculous bug caused by the dev doing clumsy ad hoc string parsing or poor null checking (more common than memory safety issues, IME)
There's generally a culture of comprehensiveness and solidity I find lacking in most mainstream programming communities
Debian says they intentionally opted not to remove these images from Docker Hub and to leave them as historical artifacts, telling users to only use up-to-date images and not old ones.
The maintainers made this decision as they believe the requirements for exploitation are unlikely, such as requiring sshd installed and running on the container, the attacker having network access to the SSH service on that container, and using a private key that matches the backdoor's trigger logic.
Idk that seems pretty reasonable to me. I think I've eojly ever needed to enable ssh on a container once
Researchers from AquaSec have noted its ability to automatically switch to backup mining pools if a primary one becomes unavailable, ensuring continuous operation. This level of sophistication has led security experts to believe that large language models or other automation frameworks may have played a role in its development.
Is it just me or is this not a very convincing rationale.
It's just a consequence of independent file formats. There's bound to be overlap in what counts as technically a valid X and also technically a valid Y. It's pretty much unavoidable. The tricky part is figuring out what fits in that sliver of the venn diagram but is also useful as malware.
FWIW, I was hesitant about obsidian for the same reasons. I would've preferred an open source editor and a syntax like asciidoc. But the fact that everything is markdown and it being such a common standard does make obsidian being closed source more palatable[^1]. And tbh, for note-taking/"second brain" purposes, a relatively constrained format like markdown is pretty suitable. I wouldn't want it for technical writing but it serves the purpose for quick and dirty tasks like quickly jotting down notes[^2]. And any other markdown language wouldn't have the same amount of tooling (e.g. org-mode is underspecified and essentially emacs-only unless you see stick to a specific subset of features)
[^1]: see the creator's blog post: "File Over App"
[^2]: in an ideal world a more sane/context-free syntax like Djot would have been nice
I understand the definition of "Freedom" as laid out by e.g. the FSF. I was explaining why your argumentation is not convincing unless the audience already agrees that complicity in genocide is an acceptable tradeoff to software freedoms. I'm saying you could make a more convincing argument by just not making that comparison in the first place. Unless your point was "perhaps we should reconsider whether Open Source is Good".
I don't know if "freedom to modify source code" and "committing a genocide" are morally comparable. This seems to undermine your point. I would have picked a different analogy
The bool_ data type is very similar to the Python bool but does not inherit from it because Python’s bool does not allow itself to be inherited from, and on the C-level the size of the actual bool data is not the same as a Python Boolean scalar.
and likewise:
The int_ type does not inherit from the int built-in under Python 3, because type int is no longer a fixed-width integer type.
my personal theory is that it's a cultural problem. same reason web apps are such memory hogs. front end web devs, as a culture, do not care about performance, but they really care about developer experience. the opposite is true of systems developers
Note that Obsidian parses all the files and uses the browser's built in database to store indexed metadata and links between note files. It's how they have a responsive search function. You could either use sqlite, ad-hoc your way with flat-files.
Of course, you could parse, walk links, and search through all files on the fly all the time. Text files at personal-notes scale should cache really well, and it might still be faster without the browser-tech overhead, but it's a toss-up
That point is when you have a human expert validate that the ai-generated code is correct. If the community as a whole has given up on doing that, it does not retroactively make all LLM-generated code not-slop, it just means slop is the norm.
And obviously from the reception you've gotten, even that has yet to occur.
Or for a more purpose built wrapper (for the same use cases as postman): https://hurl.dev/
Still relevant because claude's CLI (which is actually more of a TUI) is built on react
you're not wrong but you could've delivered that more nicely lmfao
personally, I find it useful as a crude heuristic for identifying software that probably won't make someone wake me up in the middle of the night because it crashed because of some ridiculous bug caused by the dev doing clumsy ad hoc string parsing or poor null checking (more common than memory safety issues, IME)
There's generally a culture of comprehensiveness and solidity I find lacking in most mainstream programming communities
Idk that seems pretty reasonable to me. I think I've eojly ever needed to enable ssh on a container once
SQLite continues to be the "Do Nothing. Win" of databases
In fact, that model (conceptually, though not technically) is how most fediverse software already work
Is it just me or is this not a very convincing rationale.
It's just a consequence of independent file formats. There's bound to be overlap in what counts as technically a valid
Xand also technically a validY. It's pretty much unavoidable. The tricky part is figuring out what fits in that sliver of the venn diagram but is also useful as malware.For those coming from the future: https://lobste.rs/s/aa7ske/anubis_now_supports_non_js_challenges
FWIW, I was hesitant about obsidian for the same reasons. I would've preferred an open source editor and a syntax like asciidoc. But the fact that everything is markdown and it being such a common standard does make obsidian being closed source more palatable[^1]. And tbh, for note-taking/"second brain" purposes, a relatively constrained format like markdown is pretty suitable. I wouldn't want it for technical writing but it serves the purpose for quick and dirty tasks like quickly jotting down notes[^2]. And any other markdown language wouldn't have the same amount of tooling (e.g. org-mode is underspecified and essentially emacs-only unless you see stick to a specific subset of features)
[^1]: see the creator's blog post: "File Over App" [^2]: in an ideal world a more sane/context-free syntax like Djot would have been nice
I understand the definition of "Freedom" as laid out by e.g. the FSF. I was explaining why your argumentation is not convincing unless the audience already agrees that complicity in genocide is an acceptable tradeoff to software freedoms. I'm saying you could make a more convincing argument by just not making that comparison in the first place. Unless your point was "perhaps we should reconsider whether Open Source is Good".
This assumes the audience will agree that genocide is an acceptable tradeoff for software freedoms.
I don't know if "freedom to modify source code" and "committing a genocide" are morally comparable. This seems to undermine your point. I would have picked a different analogy
From numpy's docs:
and likewise:
are there ligatures for monospace fonts that don't preserve the width of the characters?