nsonha

24 ms·

391.
by nsonha 2y ago
my workflow involves switching between files in working set, sometimes going back to staged. The git panel is easier to navigate than the file tree. I also need to see all the markers for recent changes I made, they are in git diff Inside t
392.
by nsonha 2y ago
a program can be unixy: taking inputs and producing output, and does not listen and react to any event. It can also be an UI program that are "event-driven" but all events are from user actions, hence no autonomy.
393.
by nsonha 2y ago
I have seen a few of these tools and yet to see usefulness further than not having to edit config manually for each client. Wishlist for something I would use: - Always run MCP in a sandbox - If I am gonna browse open source MCP and try the
394.
by nsonha 2y ago
There is already a definition in agent oriented programning. It has something to do with own sensors of environment and react autonomously. I find that definition fits agentic AI too. My rudimentary interpretation is anything with its own i
395.
by nsonha 2y ago
Could someone please explain in-browser inference to me? So in the context of OpenAI usage (WebLLM github), this means I will send binary to OpenAI instead of text? And it will lower the cost and run faster?
396.
by nsonha 2y ago
if it's real that person interviewed at least one candidate per day last year. Idk what kind of engineering role in what kind of org where you even do that.
397.
by nsonha 2y ago
this space is moving really fast, I suggest before forming an definitive opinion try the best tool, such as the latest Claude model and use "agentic" mode or the equivalence on your client. For example, on Copilot this mode is br
398.
by nsonha 2y ago
Stop calling HTML hypertext, it has been unstructured content for as long as I write code.
399.
by nsonha 2y ago
> Keep throwing more "tools" at it and hope to start building stuff on top of it Yes and it's working? People ARE CURRENTLY building things. They are NOT currently whinning that LLM is not smart enough so they must sit and
400.
by nsonha 2y ago
https://news.ycombinator.com/item?id=43304457 Also > ...if an LLM is failing to do this task... It CURRENTLY fails to do so, PREDITABLY and securely. What are you gonna do about that? Keep throwing more data into it and
401.
by nsonha 2y ago
1) Valid point, this could haven been wsdl/swagger. But the MCP spec supports spinning up local applications and communicate via stdio which open api cannot do. 2 + 3) having a few commands that AI knows it should call and confidently
402.
by nsonha 2y ago
https://x.com/thdxr/status/1897369381195092197
403.
by nsonha 2y ago
It's crack up to be because tools need to be hard-coded. MCP is not. When computer use was demoed it seems like a big deal. However, with MCP, any one can create and MCP server and run it on their computer and hook it up to an MCP comp
404.
by nsonha 2y ago
MCP is not that hard to understand why does it keep getting the wackiest comparison?
405.
by nsonha 2y ago
Sorry but I'm extremelly annoyed with this idiotic take that many people seem to have. Is it that easy to prompt AI to write code and call an API predictably?
406.
by nsonha 2y ago
"We should repeat the same code in multiple languages" is a pretty weird thing for professional pride. Try to be a better engineer instead of promoting gate keepers' platforms and give up "quality" as their respons
407.
by nsonha 2y ago
CSS is overly complicated for how little it does. You have to learn it. For comparison, when I did Android development, I never had to "learn" how their xml DSL works, it's mostly "just work"tm. Android has constrai
408.
by nsonha 2y ago
tldr: article that has some code and uses tools mentioned in the title.
409.
by nsonha 2y ago
My take would be starting with WAI-ARIA, or create a new standard with similar goal and keep the standard minimal and single-purpose. We are heading into the era of AI driven UI anyway.
410.
by nsonha 2y ago
from javascript? Everytime I see some function named such I assume it's something that could have been plain js but made into a hook because of amateur React devs.
411.
by nsonha 2y ago
as building blocks we of course prefer APIs. However, interfacing directly with the browser (or desktop) can enable end users to do way more things without having integration built by devs, in theory at least. In reality, LLM may not have r
412.
by nsonha 2y ago
Pretty good for a material design theme. The most boring, over used yet ugly and style-over-function design.
413.
by nsonha 2y ago
It doesn't? Freeze when commiting operations.
414.
by nsonha 2y ago
It's not like communist art is so abundant in the word that it's so boring to have another collection of them.
415.
by nsonha 2y ago
the irony is that such a style itself contributes to the diversity of art, by its mere existence
416.
by nsonha 2y ago
Maybe... quantity? Of apps an features. The idea that software is some sort of art form that should be upheld is largely dev/designer centric instead of consumer-centric. The only type of quality that matters to me is I can achieve wha
417.
by nsonha 2y ago
ikr, it had amazing camera and ran the latest linux kernel /s
418.
by nsonha 2y ago
could this just be an MCP server/ChatGPT plugin?
419.
by nsonha 2y ago
Without checking out this particular framework, the word is sometimes overloaded with that meaning (LLM personality), but actually in software engineering in general, "agent" generally means something with its own inner loop and b
420.
by nsonha 2y ago
That's just reactive programming, nothing new. It's sad that some people think it can only be achieved with wacky non-standard syntax.

More ›