Comments on: P3PC: Digg Widget

By: slamm

slamm — Mon, 16 Aug 2010 17:38:52 +0000

Kudos to Digg for responding to this review and following all the best practices now: http://about.digg.com/blog/speeding-along

By: Philip Tellis

Philip Tellis — Thu, 04 Mar 2010 00:46:08 +0000

I’ve documented how to make scripts that use document.write non-blocking here:
http://tech.bluesmoon.info/2010/01/handling-documentwrite-in-dynamic.html

By: Steve Souders

Steve Souders — Mon, 01 Mar 2010 20:05:20 +0000

@Bruno: These techniques are definitely possible, but tricky to get right across all browsers. I don’t see a working example in that blog post, for example.

By: Steve Souders

Steve Souders — Mon, 01 Mar 2010 18:26:00 +0000

@Leandro: I had a webhosting issue this morning. Should be fixed now.

By: Leandro

Leandro — Mon, 01 Mar 2010 15:39:54 +0000

The “Compare” link seems broken.

By: George

George — Mon, 01 Mar 2010 14:18:38 +0000

I do believe you’d have some fun with IntenseDebate or Disqus, Steve. :)

By: Morgan Roderick

Morgan Roderick — Mon, 01 Mar 2010 09:26:40 +0000

Very interesting analysis.

I would love to see something like this done for Omniture Sitecatalyst, so that I have you in my corner, next time I have a conversation with them about performance :-)

By: Bruno

Bruno — Mon, 01 Mar 2010 08:59:15 +0000

Hi Steve.

As you say for few times, 3rd party content are really painful for speeding up website.

Hopefully, even if they (3rd party content providers) do not want to make efforts, we can use tips for a faster impression of rendering. What’s your opinion on this kind of tip ? Do you have a better solution ?
http://www.speedingupwebsite.com/2010/02/17/removing-blocking-adsense/

By: Andrew Dupont

Andrew Dupont — Mon, 01 Mar 2010 05:01:34 +0000

@kangax: You’re right about jQuery, but recall that Digg is the site that once loaded jQuery, Prototype, and a third library (can’t recall which) on every page.

By: kangax

kangax — Mon, 01 Mar 2010 04:32:03 +0000

Curiously, first script creates 2 “variables” via undeclared assignment. This is a bad practice (for example, it will throw errors in strict mode of ES5).

The script following it includes an entire jQuery (as you have already mentioned), which is about 23KB gzipped. This isn’t a terribly lot, but considering that they only use few more-or-less trivial jQuery methods, one of the biggest savings would be to replace jQuery with a set of custom JS helpers. This would bring JS down from 23 to about 3-5 KB. And that’s a saving of ~20 KB.

The script after jQuery continues with undeclared assignments, polluting global scope with 14 more “variables” (albeit, prepended with “digg_”). It’s not clear why they couldn’t encapsulate it all for internal use of a widget.