Live data from Hacker News

Ask HN: What charts javascript library to use?

news.ycombinator.com

11–20 of 67 posts

Re: Ask HN: What charts javascript library to use?

#12

At New Relic, we use HighCharts ( http://highcharts.com/ ). We constantly get questions from our customers about which library we use, because they like the way they look and function. Highly recommended.

I second highcharts, i find it an absolutely excellent chart library. I even created a PHP wrapper for it so i could control it from the backend.

Oh and I did this because i was bored: http://zestmonkey.com/charts.php

Re: Ask HN: What charts javascript library to use?

#16
post #3

Flot is worth a shot. There may be a python library to generate is server side. When I looked into it, I kept going back and forth between javascript charts vs images. I ended up going with images because I didn't really need dynamic charts. There were also features missing from many. IIRC, Flot couldn't add a title to the x/y axis at the time I was looking at it.

Use Flot on several of my sites. They have some nice plugins that I like too, and it integrates very nicely with jQuery.

Re: Ask HN: What charts javascript library to use?

#20
We use Flot currently, but we're looking at purchasing a license for Emprise. Specifically, they offer candlestick charts, which are difficult to find elsewhere, yet absolutely critical in expressing the volatility of series data.

http://www.ejschart.com/

IMO, candlestick charts are way underutilized. Take application response time graphs for example. Most reporting suites rely on moving averages. If your application can have an acceptable average response time, but still have a high level of deviation. Let's say you have an app where, on average, one out of six requests is an entire order of magnitude slower than other requests:

  # Ruby code where r is response in ms
  r = [
    100,
    100,
    100,
    100,
    100,
    1000
  ]

  r.inject{ |sum, ms| sum + ms }.to_f / r.size # => 250
So my average response time is 250ms, which I may consider acceptable, yet some responses are taking 1000ms, which I wouldn't consider acceptable. With a simple line chart, I wouldn't see the significant deviation. With a chandlestick chart, the upper boundary would be blindinly obvious.
Post reply on HN