Archive for March, 2006

Faster DOM Queries

Tuesday, March 28th, 2006

Dean Edwards has posted a demo with a cross browser way of super quick DOM queries.

Limitations

* As described earlier, speed improvements only apply on page load for IE. Further DOM queries will use the original getElementsBySelector function
* the speed of CSS2 queries (attribute selectors) are not improved for IE5/6
* because of the nature of the CSS hack, you can only apply one behavior per element :-(

Conclusion

* DOM queries on Firefox seem pretty quick
* XPath is about 150% faster than DOM queries on a Mozilla platform
* XPath is about 1000% faster than DOM queries on an Opera platform
* the expression hack is about 200-400% faster on an IE platform
* Behaviour leaks like a sieve

Gmail Whoas

Wednesday, March 15th, 2006

Gmail and Talk have been having some trouble again. It’s been slow more often than not and I captured a few more *error modes*. Today I saw Gmail chat recover Google Talk - We're Back! with this little gem. Hey at least they recognized they are back. Seems I’m not the only one who has had trouble with gmail recently.

Gmail greeted me with this screen for a couple hours today.
Gmail Server Error

Seems now that the day is over things are back to normal and Gmail is happy. Well tomorrow is another day… we’ll see how the big G holds up.

500 Mashups

Sunday, March 12th, 2006

ProgrammableWeb.com now has over 500 mashups listed.

As shown on Mashup Feed the average new mashups per day is up to 2.81. And that is just those added to the database here. This is up from 2.6 or so just about 8 weeks ago.

So looks like it may hit 1000 before the year end.

JSONRequest

Sunday, March 12th, 2006

JSONRequest has been proposed by Douglas Crockford as a new native method for web browsers to exchange data with servers. The biggest notable difference with XMLHttpRequest is that it doesn’t prevent requests to multiple domains. JSONRequest can make requests to any server, not just the origin server like the current XMLHttpRequest. Web applications today that need to make cross-domain requests must use a proxy or a dynamic script tag to make requests to non-origin domains.

JSONRequest also has provisions for long lasting server initiated, live push of data. Crockford calls this “Duplex”.

“JSONRequest is designed to support duplex connections. This permits applications in which the server can asynchronously initiate transmissions. This is done by using two simultaneous requests: one to send and the other to receive. By using the timeout parameter, a POST request can be left pending until the server determines that it has timely data to send.

Duplex connections can be used in real time notification applications such as process management and finance. It can also be used in collaborative applications such as instant messaging, instant email, chat, games, presentation, and shared applications. “


Alex Russell
calls this same server-push technique Comet. I actually like Comet a bit more than Duplex. Duplex is a bit overloaded with a few other uses in tech. Like the half or full duplex with configuring network cards or terminal echo. Not as if I have a vote(just like AJAX sotra stuck), but let’s hope Comet wins.

Crockford on JSMin and Obfuscation

Tuesday, March 7th, 2006

Douglas Crockford knows a thing or two about compressing JavaScript. He wrote JSMin which IMHO is the simplest and fastest JavaScript minifer out there today. Coupled with his lint checker JSLint you’ve got two really good tools to write good solid code. I used JSMin to optimize digg.com’s homepage a few weeks back. As more and more AJAX apps start to take advantage of JavaScript the size and footprint grows dramatically. Even more so when folks import the various AJAX toolkits as a starting point. This means that tools like JSMin become a required part of the build/deploy process for those that are serious about optimizing end-user experience. There are other excellent minifer tools out there including Alex’s ShrinkSafe which is based on Rhino. ShrinkSafe takes an aggressive approach to not modifying the API’s as it uses a real JavaScript compiler/run-time to shrink the code. This is important when you don’t want your public API to change as is true for any JavaScript toolkit developer. Crockford makes the point that obfuscation is really just a small hurdle and any serious programmer, given time can extract the useful bits. This is especially true for JavaScript where no matter how much mangling of the code is done before it’s sent to the browser it’s still sent in plain text. So it’s trivial to attach a debugger like Venkman and walk through each line of code. Glad to see Douglas putting focus back on JavaScript as noted by his many recent updates to JSLint and now his first blog entry on the Y! UI blog. I’d also guess that Yahoo’s focus on JSON for their REST api’s is in part due to Doug joining them.

ref: Minification v Obfuscation » Yahoo! User Interface Blog