Below are random thoughts on three days in Boston at The Ajax Experience. I gave a talk titled Ajax in the Enterprise.
… AOL was giving away T-shirts by the hundreds, in solid, bright colors with words like “Geek” or “2.0″ on them. Some conferencegoers walked away with several, which wasn’t discouraged - there seemed to be an endless supply. Every time I walked by the booth I wanted (”want” isn’t the right word, it’s more the primal lust for free stuff) to grab a shirt or two, when the last thing in the world I need is more T-shirts. Eight years ago when I moved I gave 99 shirts, most of them unworn, to the Salvation Army. These were garish and ugly, and trying in AOL’s typically disingenuous way to plug into geek culture. Plus, there’s the associated guilt of all those unopened CDs taking up landfill space. Hence the little cartoon angel on my right shoulder. Of course, the little cartoon devil on my left shoulder made himself heard: “Dude, it’s FREE! Don’t be such a Herb! If you don’t take one, someone else will get it. You’ve earned it. And dude: it’s FREE!” So it went, and every walk past the AOL booth became an
exercise in self-denial of something I didn’t actually want at all, and the thimbleful of pride I felt on passing by empty-handed made me feel like some sort of lame modern-day ascetic.
… speaking of AOL (I don’t mean to pick on AOL, even the newly warm and fuzzy AOL, but they make it really easy), one of their minions gave a keynote during which he screened a movie where they went up to people on the street or in the grocery store and asked them what they thought “Ajax” was. And OH MY GAWD (the voice here should be Dr Cox from Scrubs), they thought it was a cleanser! What a bunch of losers!
Crikey.
… really regret my get-out-of-Dodge trip planning, which made me miss Brendan Eich’s closing keynote on Javascript 2.
… “stack” is the new “solution”.
… Selenium, a product for doing functional client testing in an Ajax environment, got a lot of buzz. Our QA team is looking into it. They’ve had to pound on Mercury’s QTP pretty hard to get it to work with our client, and I’m hoping Selenium will be way more straightforward.
… the booths from some of the sponsors such as Google, Sun, and Ask were more focused on recruiting than on demos, always a sign of a technology on its way up.
… after the experts’ panel on Monday night, I headed to the hotel bar and met up with the Netflix crew. As a longtime Netflix member, I couldn’t resist rattling on about what I liked and disliked about their site and service. Not only did they listen - brave souls - they bought me beer. The more beer I had in me, the more clever and insightful I became, and I think they caught on to that.
The Netflix UI makes for a great case study in how going from HTML to Ajax can improve the user experience.
… there is a bewildering array of Ajax toolkits out there, and at least two of the presentations were overviews of frameworks. The space has exploded, and sooner rather than later, it will contract. Documentation, often overlooked, will be key to adoption. It’s hard to bet against Dojo, with its breadth, careful design, and technical impressiveness. I heard good things about jQuery, and YUI looks clean. For all-Java shops there’s GWT, and possibly DWR.
… My ad hoc plan of attack, and reviews, in order of attendance:
Leveraging Ajax for Enterprise Application Development - mostly good content, somewhat awkwardly delivered. Best part was practical tips for shops coming up to speed on Ajax.
Javascript Exposed: There’s a Real Programming Language in There (Pt 1), by Glenn Vanderburg - had intended to go to Mahemoff’s talk on design patterns, but he didn’t make it to the conference. Glenn’s an engaging presenter, which eased
the fact that it was mostly a tutorial of what you’d find in the O’Reilly book. Glenn did a good job of focusing on important bits of the language that are often glossed, and I really liked his lists of what he believes the language designers got right and wrong.
Ruining the User Experience, by Aaron Gustafson - the scope of this presentation was smaller than I had hoped. From the summary, I thought it would cover UI idioms that have been made possible by Ajax but which don’t improve the user experience, but it was almost all about what to do when Javascript is disabled.
Frameworks Guide, by Nathaniel Schutta - came to this one as a refugee from a TIBCO-centric talk on building desktop-like apps in Ajax. It was good to get an overview, delivered capably and neutrally, since having developed our own toolkit I’m not familiar with others.
Intro to Dojo / Dojo in Depth, by Alex Russell - Dojo looks pretty awesome, and Alex is super-smart. Dojo obviously goes beyond Ajax, and even within Ajax it offers much more than widgets and neat effects. Their event system, the new SVG/VML support, and XhrIframeRequest were highlights. I’ll be looking soon into using its package system to leverage deferred loading of ZCS components.
IE7: From Ajax to RSS and More: How to Take Full Advantage, by Chris Wilson - acutally: quite: interesting. IE7 will make app developers very happy, and app maintainers very unhappy. It’s really frightening to think of the untold numbers of websites out there that rely on the many ways in which IE6 is broken (insert truism about reaping what you sow). As much grief as they deservedly get, MS appears to be committing to standards as far as its browser goes (thank you, market forces). Best news for me is that they fixed the garbage collector, so memory no longer leaks at the drop of a hat. They also fixed the SELECT element’s behavior with regard to z-index, the gzip caching problem, and untold CSS bugs.
JSON: The X in Ajax, by Douglas Crockford - the case for JSON is a strong one, as XML is often overkill for shuttling data around. I love its simplicity and compactness. Of particular interest: JSONRequest and a proposed MODULE tag to address cross-domain security issues. Near the end there was an odd moment when John Resig (jQuery) posed a reasonable-sounding question about whether JSON performance scales well across large datasets when compared with XML, and the immediate response was “Next question.” Not sure what to conclude from that.
Designing for Ajax, by Bill Scott - bailed on a performance talk that turned out to be Gomez-centric and came here. Great presentation. Bill used Yahoo as well as other sites to point out good and bad Ajax-enabled interaction patterns. Makes me want to check out the Yahoo design patterns library.
Case Study: Building Great UI, the Netflix Way, by Sean Kane - everyone loves Hypnotoad, I mean Netflix. What was interesting here (aside from the obvious fascination with the service) was how extensively they test in UI labs with regular folks before rolling out changes, often trying out several possible versions of a UI feature. It would be nice if we could eventually get real-world
data like that.
Last week I posted an entry comparing IE 7 with Firefox 1.5 and the venerable IE 6. Lot’s of folks have since asked for Firefox 2 to be added to the mix. The reason we didn’t initially include Firefox 2 is because it was not at the time officially released and so we felt it unfair for it to be included.
The graph below shows the cumulative time it took for IE 7 and Firefox 2 to execute the Zimbra Web Client (ZWC) peformance tests at various dates leading up to the final release of each browser. As can be seen, both IE and Firefox showed improvement in their final release, and so we think leaving Firefox 2 out of last week’s tests was indeed the fair thing to do.
So now that Firefox 2 has been officially released, how does if fare against IE 7?
We ran each browser over a common set of Zimbra Web Client (ZWC) operations such as logging in, viewing messages, navigating around various folders, changing options, viewing contacts, and performing various calendar operations. As the graph below shows, Firefox 2 beat out IE 7 in just about every operation - sometimes by significant margins.
We also tested Firefox 2 vs Firefox 1.5. As seen below, Firefox 2 has not made substantial performance improvements over Firefox 1.5, and has certainly not shown the same order of gain as IE 7 has over IE6.
I think the net result is that for highly JavaScript intensive AJAX applications such as the ZWC, Firefox is still the winner by quite a wide margin. Of course your mileage will vary depending on the type and footprint of your client.
I would like to end by saying that we are able to present this data thanks to lots of hard work by Raja Rao of the Zimbra QA department who has spent many long hours building our AJAX client testing framework, and executing the performance tests against the various browsers.
Back in April I wrote an entry complaining about IE’s performance as a Web 2.0 platform:
“From a Web 2.0 application developers perspective (developers who use a lot of JavaScript and DOM manipulation), IE 6 is plagued by a number of well known problems such as its ability to readily leak memory. Regrettably, Microsoft’s next release of Internet Explorer, IE 7, does little to resolve these issues.”
I am happy to say that I was wrong.
Microsoft’s IE team has clearly been hard at work on improving their browser’s memory management and JavaScript performance. IE 7 has made some significant leaps forward based on some initial in house testing here at Zimbra. We are in general observing about a 2x performance improvement with IE 7 vs IE 6 when using the Zimbra Web Client (ZWC).
As is well documented, IE 6 is notoriously bad at leaking memory, particularly due to circular references that include COM objects. The good news is that our tests indicate that this problem has been solved in IE 7. While in our test profile, it appears to consume more memory than Firefox, IE 7 seems to have solved the horrendous memory leaks exhibited in IE 6.
We also looked at the performance of Firefox, IE 6, and IE 7 over a set of common ZWC operations such as logging in, viewing messages, navigating around various folders, viewing contacts, and performing various calendar operations. The graph below shows the relative performance of each browser against the other:
Again we see that across just about every operation, IE 7 performs better than IE 6; however, for the most part Firefox still beats out IE 7. When we looked at the sum total time it takes for all operations to be performed (admittedly a coarse grained metric), we noticed that IE 7 was about twice as fast as IE 6; however, Firefox was more than twice as fast as IE 7 and about four times faster than IE 6.
In conclusion, IE 7 has made some quite significant improvements over IE 6, both in terms of performance and memory management; however, there is still room to improve - particularly against Firefox, a challenge I hope the IE team will be taking up.
ZCS 4.0.3 includes fixes for 55 bugs and enhancements. It has significant improvements in calendar behavior and fixes a bug in Trash folder viewing that was troubling many folks. It also includes upgrade speed improvements. This allows the upgrade to take much better advantage of installs that have multiple disks available to MySQL.
Zimbra - Forums - Zimbra Collaboration Suite 4.0.3 Released
(FYI: I’ve made some enhancements/corrections based on Zimbra’s participation at the Office 2.0 Conference; pointer below. May or may not be worth a quick reread.)
Yes, the “2.0″ hype is getting out of hand. However, just as with Web 2.0, the technology evolution we are participating in is sufficent to at least justify the discussion. So while I am still dubious about the Office 2.0 moniker, there is no doubt that the Web authoring, sharing, and collaboration technologies under the Web 2.0 umbrella are allowing us to do many of the things we used to do within proprietary Office 1.0 desktop applications, and to do so from any browser on the net. So before you dismiss Office 2.0 as yet another buzz word du jour, please consider some (modest) over-generalizations:
Office 1.0 Users: Power users (information workers/professionals)
Office 2.0 Users: “All hands” (most everyone who browses)
1.0 Model: Client-centric (desktop applications)
2.0 Model: Network-centric (web applications)
1.0 Sharing (in the small): Pass by value (Email), pass by reference (Public folders)
2.0 Sharing (in the large): Adds the ability to use the Internet to pass by value (Email, VoIP, IM, …) and (effectively) by reference (XHTML pages/hyperlinks, Wiki, Blogs, iCalendar/CalDAV, iTasks, WebDAV, …), all with access control
1.0 Navigation (sans meta-data): Folders
2.0 Navigation (with meta-data): Hyperlinks, indexing & search, tagging, mash-ups …
1.0 Editing: Via proprietary desktop applications; WYSIWYG with change tracking
2.0 Editing: From any browser; WYSIWYG (via Ajax authoring) with versioning and history
1.0 Data types: Proprietary
2.0 Data types: Open (XHMTL, ODF, microformats, XML such as via Service-Oriented Architecture/SOA, REST, SOAP, etc.)
1.0 Content: Relatively static, with intra-desktop dynamic components
2.0 Content: More dynamic (including web application-generated content, SOA mash-ups, …)
1.0 Multi-document: Object Linking & Embedding (OLE), Bonobo, …
2.0 Multi-document: Hyperlinks, Ajax Linking & Embedding (ALE), mash-ups, …
(No doubt the above takes a rather expansive view of Office 2.0, but then again Office 1.0 arguably covered most all computer-assisted authoring of content. Please also note: The above content would have proved substantially more compelling in an HTML table authored, say, with a WYSIWYG Ajax editor, like that included within the Zimbra Documents applications.)
All this is not to say that Office 2.0 can or should supercede Office 1.0—Zimbra spreadsheets, for example, will not support pivot tables any time soon. I believe our goal should not be to reproduce Office 1.0 functionality on a Web 2.0 platform, but rather to realize an easier to use (a.k.a. less complex) collaborative authoring and sharing model that scales with the Internet. Office 2.0 users should be able to (1) author content from anywhere; (2) appropriately reuse and adapt (mash up) content (both static and dynamic) already published on the Internet; and (3) securely collaborate with others in whatever ad hoc fashion best meets their needs. The fun with “Office 2.0″ is that thanks to the maturing of the underlying Web 2.0 technologies and near universal success of the Internet, this vision is close to realization.
One thing I’m not yet happy with is that the above description does not draw a very compelling line between Web 2.0 and Office 2.0. Some technologies are more obvious: for example, Ajax, RSS, and SOA/XML fit under Web 2.0, while Ajax authoring and ALE arguably fit more naturally under Office 2.0. Given the emphesis on collaboration and authoring inherent in the Web 2.0 definition, I think these two potential categories are going to be tough to keep separate. After all, both Office 1.0 and Web 2.0 are about authoring rich content—the difference is that the former is focused on the desktop and “sharing in the small”, while the latter is targeting the network and “sharing in the large”. Office 2.0 then may merely mean leveraging Web 2.0 technologies to do many/most (?) of the things we used to do in Office 1.0.
(Some of this thinking came out our our panel session at the Office 2.0 Conference in San Francisco, and our Web 2.0 Kongress session in Germany the day before. Just doing our bit to keep the airlines in the black :-).)
Zimbra Collaboration Suite 4.0 Webinar
When:
Thurs, October 5th at 10am PST. Length: 60 minutes.
Subject: Modern Collaboration with Zimbra Collaboration Suite (ZCS) 4.0
Overview:
Fueled by demand for better productivity, the web is exploding as the primary communication medium for the modern organization.
ZCS leads the way, providing powerful next-generation messaging and collaboration.
Join our CTO Scott Dietzen for this free live webinar to see:
* How ZCS 4.0 is the right solution for today’s collaboration demands
* Discuss and demo features such as Zimbra Documents, Admin UI, Zimlets, Mobile device sync, Active Directory integration and more…
* Why ZCS 4.0 enables organizations to realize the true benefits of an innovative and cost-effective collaboration solution.
* Plus live Q & A
Sign up:
Please register; space is limited!
| 
