I'm working on a visual description app for the blind. even Gemma 4 E2B can give very useful image descriptions while at the same time taking questions as audio. It's also much faster than most of the current popular cloud based apps like Be My Eyes.
This is the new challenge. Every vibecoder is larping as a SWE and rewriting systems left and right when all they need is a new feature or two. 6000 line PRs that could be 60. The challenge in the current era is evolving existing systems to ensure they meet previous requirements in terms of capabilities and performance. This is the new role of the SWE. Same as the old role, but with less typing.
My 12 y/o daughter recently ran into a "does it run DOOM" reference in media (I think a graphic novel-- not sure) and asked me about it. I got to explain the phenomenon and show her some examples (she found the pregnancy test to be particularly amusing). I'll have to show her this one.
good advertisement now to shift the tide back to openai that just works and honestly codex with gpt 5.4 is _surprisingly good_ currently, not nerfed or forgetting half the tasks along the way so far. Opus already got worse than sonnet last weeks beyond just crazy token costs, now reliabilty goes to shit and anthropic seems like using it. Meanwhile, delightful of the codex desktop app in fact, stuff seems to "just work" elegantly with good quality.
Yes, and? I wasn't trying to demonstrate jj superiority. I was responding to a post about a normal messy workflow and showing how to handle it in jj.
If I'm writing a description of how to use jj, I could take several different approaches. Am I writing for a git novice? A git expert? An expert in a different VCS? A novice in any VCS? And even within those, there's a big difference in whether you're a solo dev working alone on their own project, a solo dev working across multiple systems, a random github contributor working alone against a github repo, a group of contributors who work together before landing something in an upstream repo, or whatever. And then, it matters whether my objective is to show that jj is somehow superior, or to just show to accomplish something.
Those are going to require rather different approaches. I was not going for "jj is better than git". I was aiming more for "here's how straightforward it is to do the sort of stuff you're talking about". Even with the example actions I described, jj does have a couple of advantages that I didn't highlight: first, your git equivalents would require looking up commit hashes whereas in jj I tend to remember the recent change ids that I've been working with. Second, `jj undo` (and its stronger variant, `jj op restore`) is easier and simpler to work with than the reflog. A longer example would have demonstrated that, but I didn't want a longer example.
But I have no dispute with your assertion that this workflow is not harder in git. I could write a description of why I think jj is better than git, it's just that my post was not that. (I could also write a post about how jj is still missing some important functionality that git has, and therefore git is better than jj.)
But just to sketch out what I would use if I wanted to make git look bad, I'd probably use an example of multiple independent lines of development where I want to work off of a tree with all of them applied at the same time, without artificially linearizing them because I don't know what order reviews are going to come in, and then doing fixups and incorporating new changes that turn out to conflict, and not getting stuck working through conflicts in the other patch series when I'm actively working out the finishing touches on one of them that turns out to be high priority. And then getting some of that wrong and wanting to back up and try a different path. All while carrying along some temporary logging or debugging changes, and perhaps some configuration changes that I don't ever want pushed. And keeping my patch series clean by separating out refactoring from changes, even when I actually do bits of that refactoring or those changes out of order. And doing all this without risking modifying anything other people might be using or building off of, by preventing force pushes on stuff that matters without preventing it for in-development stuff that is only relevant to me. And in the middle of this, wanting to look back on what the state of things was last Wednesday, including the whole commit graph.
All of that is possible with both git and jujutsu. In practice, I wouldn't even try much of it with git. Perhaps I just suck at git? Very possible. I'm better with mercurial, but I wouldn't do a lot of that there either. I won't say all of that is trivial with jj, but some of it is easy, all of it is doable without thinking too hard, it's the sort of stuff that arises for me quite often as I'm working, and none of it requires more than the same handful of commands and concepts. I know what changes are tentative and what are more fixed without juggling commits vs staging vs stash, and I can freely move bits and pieces between them using the same set of commands. I could do the exact same things in git, but I wouldn't. The core git data model is very nice, so it's pretty clear what can and can't be done. jj gives me the ability to manipulate it without tangling my head or my repo in knots.
because Meta's lobbying has been publicly identified. When the other companies are found to be spending millions of dollars to push these age verification laws, then they, too, will be harped on.
Germany’s Federal Network Agency is aware of stability issues of the German electric grid on days like Easter Monday and insufficient deployment of smart meters.
How do you think we got the grifting government? It is because people were upset with what the democrats were doing. Maybe if they stopped doing that they wouldn’t have lost. They’ve failed to lean and since Trump is so bad they still won’t need to learn and will continue their mistakes and keep losing.
I ended up watching Bicentennial Man (1999) with Robin Williams over the weekend. If you haven't seen I thought it was a good and timely thing to watch and is kid friendly. Without giving away the plot, the scene where it was unloading the dishwasher...take my money!
I mean, there's truth on both sides here. Essentially LLMs are good at some stuff bad at others. If you don't use the tool appropriately you are going to hurt yourself. And at the moment there really isn't a very good opinionated good workflow framework for early adopters. Which means unless you have senior engineers the results can be varied.
I can tell you it's saved the day for me on a few occasions and lord knows it can parse docs better than any human and give you what you want in a clear cogent manner. But it's sometimes flat out wrong, and it's sometimes making things harder than you want, or isn't very context aware no matter how much you try to point it in the right direction.
But it's still good at stuff and saves time.
Early adopter pain is early adopter pain.
What scares me more is we are trading short term gains for long term losses here. And we are destroying the feeder system for senior engineers... and that's going to end horrifically.
I stated that I wouldn't advocate for a national prohibition on guns mostly because I think it wouldn't work and I support gun rights. But if you were to try banning guns, it'd have to be a huge national effort and not just a few states.
I've built something similar to what you described: https://kopai.app
OTEL is the common denominator for all the observability purposes including the ones that come from eBPF and the application layer.
Same architectural idea as Ingero. We skip the aggregation layer. We also skip the MCP part. The agent uses a CLI instead, which keeps it composable with any coding agent workflow.
You can have your claude poke at the logs/metrics/traces using a dedicated cli tool paired with some agent skills. It is Apache-licensed and you can try it locally.
I'd argue that this should be refined to something like "farmers that speculate heavily struggle in an under-regulated free market".
Financial stability in highly volatile markets depends on appropriate planning, saving, and distribution. I say this from the investment perspective, but I would venture to guess that it also applies to hard goods like food-stuffs.
The cynic in me says the Win 11 "you must have a TPM" push (along with passkey's "big tech owns all your accounts" design) were rammed through specifically to centralize control of the open web.
At this point, if the federal government actually forced OS-level censorship, most literate folks would just download Linux. So, first, they need to close the remaining door.
I'm not exactly holding a candle for debian. SystemD has already started adding support for this, and, in the past, downstream has been able to force unpopular debian votes through.
I learned recently that Spain uses the same timezone as Germany (GMT+2 currently, according to Google) despite the GMT line passing through Spain. I've visited Spain and did feel like we ate late, but the timezone being "wrong" has made me wonder if it would have felt as late if I knew about their timezone while I was there!
As I mentioned previously, the writing is on the wall. It is a matter of time.
We definitely need a true alternative on the market, preferably open, to balance things out and to free everyone from the duopoly. The political pressure that is needed is not to “keep” Android open, but to ensure that governments and institutions don't double down on the existing duopoly. Ensure that interoperability standards are in place, and don't lock people into the existing big tech platforms/solutions.
