Show HN: Searchkit – React components for elasticsearch | Hacker News Reader

Show HN: Searchkit – React components for elasticsearch

searchkit.co

74 pointsjoemcelroy27 comments

Nice! Funny is that Algolia released something very similar some time ago: https://community.algolia.com/instantsearch.js/

It seems searckit took great inspiration from it:

- component naming (search box, refinement list, hierarchical menu)

- documentation organisation:

+ http://searchkit.co/ is using an imdb demo just as https://www.algolia.com/ homepage

+ https://ssetem.gitbooks.io/searchkit/content/ vs https://community.algolia.com/instantsearch.js/documentation... (see "components" menu on the left)

- react based (but only works if you have react though)

- and more parts maybe

@Author could you give some more details on the parts of instantsearch.js that were inspiring to your project?

Thanks

Yep we did take a look at Algolia components. They made a great library and we took inspiration from their API. The difference with us is we focused on building with elasticsearch and we use declarative based components focusing towards the more experienced developer crowd.

thanks for your honest reply, good luck with the library!

I'm enthused, but would suggest you don't put an entry in the user's history for every keystroke in the search box.

Yes, this! People seem not to know when to use replaceState over pushState. And in my eyes this is a clear candidate for replaceState. (There are also examples the other way around, such as GitHub's PR tabs, drives me insane every time.)

Yes! Thanks for noticing a side effect of the way we update our url. It should be replaceState for the search box. Other actions like pagination, applying filters and sorting should be pushState however.

We also throttle the searchbox search requests as you type.

fixed now :), try the demo now http://searchkit.co

thanks, will fix this today :D

Hi I am the Co-Author of Searchkit

We have now switched to Apache license Due to interest and our wish is for many new contributors to add to Searchkit's ecosystem + component library https://github.com/searchkit/searchkit/blob/master/LICENSE We will be working hard in the next weeks to add to the developer documentation, and walkthroughs on making custom components. Thanks!

(Author) This is our first release of searchkit. Adding more components and documentation in the next few days. Let us know if theres something missing you like by opening a github issue. Thanks! Joe

Have you considered a more permissive license, like Apache/MIT/BSD?

We have switched to Apache license Due to interest and our wish for many new contributors to add to Searchkit's ecosystem + component library

https://github.com/searchkit/searchkit/blob/master/LICENSE

We will be working hard in the next weeks to add to the developer documentation, and walkthroughs on making custom components.

Thanks!

What's wrong with the GNU license?

Assuming your question arises from a state of ignorance, rather than from rhetorical flourish:

What distinguishes GNU licenses from more permissive licenses is that the GNU licenses require derived works be released under an appropriate GNU license as well.

This is not quite anti-commercial, but does put the kibosh on some otherwise viable business models. If you wish you could incorporate some of this code in a proprietary product that you can sell licenses to, then what's wrong with the GNU license - from that stand point - is that it forbids that.

That's the intended purpose of the GNU licenses - to promote software that respects user freedoms by making it easier to write such software (because of access to existing GPL/AGPL code) relative to the ease of writing software that infringes on those freedoms.

As a necessary side effect, it is also the case that you cannot move GNU-licensed code into a non-GNU-licensed-but-compatible codebase without changing the license to an appropriate GNU license. Code can move the other way, from compatible permissive license to GPL.

Sort of the worst case, in terms of unintended consequences, is code under an incompatible-but-otherwise-free license, where code can't move either way for not really any good reason.

Fortunately, people seem to have generally settled on GPL-compatible licenses, whether permissive or copyleft.

> Assuming your question arises from a state of ignorance, rather than from rhetorical flourish

It is quite clear to me that the parent's intention was to challenge the ritualistic "Why not MIT/BSD?" type of argument. It is the author's prerogative to decide which license the software should be under. While I use permissive licenses for my work usually, I see the merit in GPL licenses and might even switch back to it at some point. They serve different goals, and I suspect that in future we're more likely to converge towards GPL-like licenses.

GPL protects the rights of independent developers better than permissive licenses; and as software development becomes more accessible, we're likely to see more work pocketed off by companies without giving due credit, or even entirely forked off. GPL clearly recognizes that companies will not give back unless required by the law. If we're to have a fairer and more decentralized (political) power distribution, we need the protections of the GPL.

If I could flick a switch to turn all OSS projects to GPL, it'll impact the entire industry in a good way. OTOH, changing everything to permissive licenses will have absolutely no impact.

'It is quite clear to me that the parent's intention was to challenge the ritualistic "Why not MIT/BSD?" type of argument.'

It seems plenty possible, hence my explicit note that I was assuming otherwise. It doesn't seem clear - I've encountered people asking that question honestly and I tried to give an honest answer. To the degree that I'm a partisan in this discussion I am pro-GPL, but I think it's worth acknowledging the legitimate points raised by reasonable people on the other side.

Just a minor nitpick, but adding a GNU license to the code base do not mean you have to change the license. The distribution of a code base that combines different licenses must simply follow each and every condition of the compatible licenses.

The effective license will of course look very similar to the GNU license, since following the GNU license conditions will most of the time also mean that you are in compliance with the permissive licenses conditions, but there are exceptions. For example, GPLv3 do not forbid distributors to use the authors names in the promotion of products, but it is compatible (under the additional terms section) with 3-clause BSD that explicitly do forbid such promotion.

Good catch, thanks! Of course, the set of limitations that can be imposed by other licenses while still being GPL compatible is quite small.

Totally asking out of ignorance here: How does GPLv3 play with JS apps?

I'll normally create an app bundle that includes all the libraries, and the users will only execute that one file. Does that count as distribution? Would that make it such that the whole app must now be licensed under GPLv3? I make source maps available in production (that get dynamically loaded when you open dev tools), is that enough to meet the requirement of sharing the source?

Heck, even if you split off the GPL'd files from the main app bundle and it's loaded dynamically, you'll still be executing it on the same DOM... So it's still eating up the rest of the app into becoming GPLv3, right?

Good documentation! :)

Thanks will! We listen :D

Looks nice. I'd prefer higher-order (wrapper) components to extending the component classes though, HOCs are more idiomatic.

Another question, somewhat off-topic. How do people feel about talking to elastic search directly (via proxy or not), versus going through another service layer. The argument for a service layer is providing a cleaner and more stable API for front-end developers. The downside is that elasticsearch has a lot of useful features, and it's hard to imagine an API that's much simpler that exposes them in a useful way, you'd end up significantly slowing down development of front-end search features.

My reason for asking here, is that libraries like this are appealing, but may not fit well with the way more micro-service-y teams approach search.

Thanks, we can look into higher order components for wrapping and overriding blank states etc.

The service layer is purely opt in, and its a thin proxy incase you want to apply permissions, it speaks elasticsearch purely, this is one thing we don't want to change. E.g. our demos actually connect direct to a read only elasticsearch instance

> I'd prefer higher-order (wrapper) components

Wow. Thanks for that name. I found this solution when I needed to pass some async stuff to React component. But I called it hacky wrappping function. Now I can call it higher order component, give it capital letter name and be proud. :-)

Looks great. But don't add every key press to the browser history. Not cool ;)

thanks for the feedback, this is fixed now :)

Looks awesome. We are building something on react and we will probably use this.

thanks.

Reply on news.ycombinator.com