I almost never use npmjs.com.

When I do it’s just to click on the repo link.

I just do ghub.io/<package-name>

It is really annoying if you have a package that is relatively new to the platform, and you type in the exact package name, that package is not reliably the first result.

Minor edge case, but infuriating if you want to check your own packages quickly (without needing to navigate menu > packages > YOUR_PACKAGE).

Still agree with you though, who is npmx actually for?

Whenever I know the name of the package I want to see, I always type in the URL directly: npm.im/[package-name]

npmjs search is very slow

how often does anyone use it though?

I mean are we really arguing over milliseconds here? I have never in my life had the thought "NPM search is too slow, I need a faster solution"

I have had the thought "NPM search sorted by downloads this week is giving me irrelevant packages" - but I'm not sure this tool solves that.

It can load for a couple of seconds (!) for me

My only pain point, that I solved with a few lines of elisp, was to go to npmjs, to find the repo link for my dependencies. What's in `node_modules` may be compiled to an inscrutable blob and it's rare to find good library docs.

Life-changing: ghub.io/<package>

No idea where I found this but I’ve been using for many, many years.

How's it life changing though? You have to know the package name already to do that. And if you're sitting on npmjs looking at the package already, then you're one click away from Github.

Furthermore... I don't host my packages on Github :-) Don't think that even works for mine.

When you see an unknown package in a package.json, you can go straight to the source - no search or visiting npm.com

Cynically, if you can attract a representative sample, you could aggregate and sell analytics data.

Another could be to have an "alternatives" section based on semantic similarity and / or some other features that have signal.

Because they wanted to and Claude didn't tell them not to. Why even ask questions like this at this point.

The prolific inventor’s dilemma. But to be fair developers have been making whatever they want since the beginning. Sometimes there doesn’t have to be a ‘why?’.

In the age of LLMs I think we are going to see a Cambrian explosion of software.

Me personally, I’m writing tools for myself wouldn’t have bothered with before due the the time investment needed.

> In the age of LLMs I think we are going to see a Cambrian explosion of software.

From Wikipedia: "Application software _runs_ [1] on top of the operating system"

[1] emphasis mine.

Haha this. I understand the appeal of "but I can probably roll my own" and "Oh that's a good idea, let's jump into coding" but pre-vibe bots, the effort required would make folks stop and think before jumping into it - is it really a good idea? should I do it just because I can? what problem is it really solving? Who will use this? etc.

The website has an about page that explains some of the reasons why this project exists. The first heading in the readme of their repo is "Vision". The creators of this project are people with a track record of creating popular, high quality, useful Javascript projects.

I'd say the probability that some thought has gone into this project it pretty high. Your reply stating that this was created without any though or effort is, ironically the least thoughtful, laziest and least useful response possible.

Check who the maintainers are. This is highly unlikely. If you want to know what Daniel thinks of Claude read https://roe.dev/blog/the-golden-thread

It sparks more joy than the old one and buddy, that's for frelling enough damn it. Whinge out!

Awful comment. Your comment is bad and should be ashamed.

Use it and disagree! Tell me it in fact does not spark more joy! This just seems pretty clear, ya'll.

It's just generally vastly nicer. I love that file exploration of packages doesn't feel like a last afterthought before leaving the solar system forever.

https://github.com/npmx-dev/npmx.dev has a comparison

npmjs.com search input doesn't even have proper throttling/debounce. If you type fast, there is a high chance that you get result of the partial input that you entered than the complete one. Ideally they should discard the response for the old partial search if there is a new one.

A valid use case I can think of is if npmjs.com is blackholed by your company to prevent supply chain attacks

When I checked the site yesterday after reading this, I found that npmx was indeed faster. And on npmjs.org, the back button was broken when browsing code on the site, here not.

However, I would be wary to adopt a third-party code browser for previewing dependency code, even more than using the auhtoritative one on npmjs.org.

Downloading an archive without executing anything and looking at the content would be better if done for getting a peek.

I don't get the strong dislike for the UI though, it looks solid to me. Aesthetically conservative and nothing new, but well-executed, responsive, tidy, and with attention to detail.