Sergey Chernyshev – Page 14 – Binary Orders of Magnitude

Performance @ New York Web Standards Meetup

Yesterday, I talked about Web Performance at New York Web Standards Meetup hosted by theMechanism.

Thanks to all attendees for coming to the presentation, I hope it was informative and worked as a call to action.

If you have any questions or need help with your performance optimization project, feel free to post your comments here or contact me directly if you want, I’ll be happy to help.

HTML Slides with meta-data embedded using RDFa are available here

If you’d like to watch more presentations on performance (already have about 30 of them), you can visit TechPresentations: http://www.techpresentations.org/Performance

Including a couple presentations about YSlow and related tools: http://www.techpresentations.org/YSlow

Once again, thank to all attending, asking interesting questions and volunteering for YSlow treatment ;)

Sergey

TypePad Connect and rel=”canonical”

Some time ago I created MediaWiki Widget for TypePad Connect:
http://www.mediawikiwidgets.org/TypePad_Connect

While editing a wiki page and going to different versions of it, I noticed that Comments widget has a well known issue – it has no idea which comments should be displayed on the page if page has different URLs (ad campaign tracking codes will probably be a main commercial issue, but also editing interfaces in MediaWiki and other CMS softwares and so on) – you can see example of what I’m talking about on these two pages:

which represent the same content page with different URL.

Other frameworks like JS-Kit, for example, solve this problem by allowing to pass URL or ID of the page to the widget, but I just had an idea about new approach to this issue.

The thing is that Google, Yahoo! and Microsoft just announced a new tag in order to fight SEO spam (and I already installed MW extension from Wikia):
<link rel="canonical" href="http://www.mediawikiwidgets.org/TypePad_Connect"/>
The idea is very simple – they just need to know all the pages that legitimately think that they are copies of page X so their smart algorithm that fights page duplication spam will have more clues.

But the side effect is that tools like TypePad Connect or JS-Kit can automatically, on the fly identify the pages they should serve content for. All you need to do is to write a simple jQuery code like this (my first line of jQuery) :
var canonical = $('html:first > head:first > link[rel=canonical]').attr('href');
I think all of us will benefit if widgetized tool providers like TypePad will push for this as solution to the page identity problems.

I hope this will be included into a list of high priority issues and TypePad Connect will get improved first (I actually enjoyed creating widget for it – it was much-much easier then with other vendors).

New OpenID look for MediaWiki

I’ve done some integration of OpenID selector into MediaWiki OpenID extension and some fixing/improvements of the latter.

You can see changes on TPr, MWW and SharingButtons.

Next week will take a look into checking it back in. Still need to figure out if I can check in BSD code (selector) into MW tree or need to write something similar (I’m even thinking about creating something like SharingButtons, but for OpenID providers) and release with GPL.

OpenID and memcached

Did some maintenance on my projects – added OpenID support to wikis:

Also working on embedding OpenID selector – check Sharing Buttons OpenID login page (still needs some work, but quite promising).

Also changed all wikis and blogs to use memcached instead of APC for cache storage – it seems to be much more efficient.

New name

While walking to work early today I came up with the name for my blog – “Binary Orders of Magnitude”. I think it’s a good name for personal blog and sounds very big and global as all blog names should sound as all of us live to accomplish something big. Plus it has some geekish humor to it ;)

Hope you enjoy it – let me know what you think.