Fullproof - A javascript fulltext search engine library | Hacker News Reader

Fullproof - A javascript fulltext search engine library

kornr.net

88 pointsd0ugal18 comments

What is the motivation? Adding search functionality to sites that have no dynamic backend?

How would I build an index? Are tools included for that, or is it just reading an index? Is it building the index each time the page is loaded, or in advance, or is both possible?

While the slides explain, what a search engine is, they seem not really to explain, how I can use this particular search engine, which is a pitty, since this looks like a cool project.

This is the project page, but I cannot find such information there as well: https://github.com/reyesr/fullproof

I am working on a project which is, essentially, fully client-side. We decided to use the browser's built in search function to handle searching until we could implement a better search method. We also have tagging in place because that's how we handle our most basic of searching. Having a fulltext search library that was would allow us to search through our data more effectively. Now to test this on the mobile browser.

I would think that even for a static site you'd at least want to generate the index as part of the build process, not (like fullproof) do that on the client as well. Of course, for client-side apps that never touch a server and store user content in Local Storage or some-such, that's not really an option. I'd wager 99% of us don't have that problem, though. Still, kudos to the author, looks well thought-out.

Forget fulltext search, I'd like to know how he pulled off that 3d cube scrolling! *Found it: http://lab.hakim.se/reveal-js

I have to concur, the visual side of the presentation is quite nifty and appealing.

Yes, but it breaks usability slightly - it overrides the back button.

I cannot say it's usability break if using back button allows me to go to the previously seen slide and after that forward button brings me back where I was. The only problem is that you have to wait a bit until URL is changed. OTOH it is quite useful too, because quick scrolling over the slides isn't recorded in the history, only the slides where you were for more than a mere second or so.

Not to mention that it doesn't scale well with screen size (I can't see the bottom below tokenization on slide where it shows structure). It looks nifty, but that's about it.

While we are on the topic, you should check out some of Hakim's other work. Meny and Forkit in particular are two really cool scripts.

Forkit: http://lab.hakim.se/forkit-js/

Meny: http://lab.hakim.se/meny/

Wow, thanks

The presentation reminded me of ancient microfiche readers...

you should see impress.js.. thats another good one http://bartaz.github.com/impress.js/#/bored

Nice idea, but how do you get around storing and retrieving large amounts of data?

in-browser databases (syncing with server) like pouchdb might take care of this: http://pouchdb.com/

A bit off topic. I FUCKING HATE this new 3D navigation shit that some websites started doing. It's slow, it jumps through pages sometimes, it's absolutely not clear where you are.

New? This tacky crap has been around forever ;)

http://desk3d.sourceforge.net/screenshots.php

The difference is that it was kinda cool as a way to browse between desktops (and apps, to an extent.) But for browsing a stream of continuous information it sucks. There is no easy way to fast-forward/rewind to a specific point, etc.

Might work for a short presentation/tutorial form, which is what I think this site is trying to do.

It's a really horrible interface for technical information. Used as a viewer for a handful of images might be bearable. But for describing a javascript implementation of a full-text search engine? No thanks. Give me black text on a white background and a couple of graphs.

Reply on news.ycombinator.com