[Ask HN] Review my weekend project - linkloo.com | Hacker News Reader

[Ask HN] Review my weekend project - linkloo.com

linkloo.com

18 pointsadrinavarro15 comments

A very simple idea: pick 50 main or interesting sources, build a stack coupling the whole thing with superfeedr and some scripts for sphinxsearch, a database and a nifty frontend.

Search example: http://linkloo.com/#twitter

Why? Well, when something important happens (twitter xss, ipad update, pope visit…) I usually go through main sources looking for information. This way, I just have to write a search query and get some interesting insights.

And near real-time. The homepage is a real-time stream, the search results don't update in realtime... yet, but results are also fresh (delay is under 20 seconds and superfeedr is pretty fast).

Tips? Useful for anyone or not worth it? :)

[edit] added more ram to the instance and tweaked fcgi settings, forgot to do it before, should work faster now

Interesting! Make an iPhone friendly version :)

I'd recommend putting somewhere on the page 'instant news search' or hopefully a better tagline.

I had no idea what linkloo was until i read the page title which is partially hidden in the tab.

nice idea, not sure about the name though, I'm English and loo means toilet to me. I assumed it would be lots of 'crap' links not good links.

Crap, didn't really think about the name, I just had the domain name lying around (was the name of one of my first projects, when I was 13 or 14).

For now it's just a concept. Maybe if it ends being something useful for people on a daily basis and I can get enough time to push interesting features I'll change the name and establish it as a serious project.

Good idea, great execution.

How do you envision the site being used in the long term, and do you think there is a way to monetize it?

Being honest, nothing at all. And it takes ~25€ to run it each month (superfeedr credits, virtual machine).

Honestly I need some good tips to make this more than a occasional tool. But I definitely don't want to be a replacement for google blogsearch nor technorati, but rather be an interesting 'hub' for contents (and realtime, that too).

This could be actually be a good way to recommend new articles to people. Have suggested articles on the side, based on previous search terms. This could become pretty accurate as the user logs in over time.

Interesting idea. I definitely see this as something I might use on a more regular basis.

There's one annoyance, though: instant search generates a lot of "browsing history", so I couldn't get back to HN with just one click on the "back" button. At least that's what happens with Firefox 3.0.19 on Windows.

What about maintaining a list of "last 24 hours" "most searched / mentioned in the news" terms for them to autocomplete via jQuery in the search box , a la Google Instant ?

That way you could get informed about popular terms in the last 24 hours you didn't even notice.

It's great :) Love the instantization of select news sources. Maybe you can provide an on-the-fly custom selection of news sources - or adding custom news sources by RSS

The site is good, idea is fresh. Just one thing, search needs more work, I will suggest you make regular search work first before going instant.

For example? The search is powered by sphinx, the title, summary and tags are indexed and words are stemmed. Search is fast, instant search is error-proof for concurrency problems (or at least it should). There is no synonym search, but this isn't google either.

Sources button is broken in Chrome (running Ubuntu 10.04 derivative).

Not for me, exact same setup.

Reply on news.ycombinator.com