Blogging Roller

Netflix is raising prices again, but Kuro5hin suggest that by attempting to cancel your Netflix account, you can get a better deal - even better than before the price hike. I've been a Netflix member for five years now, so I remember how good the service used to be. Turn-around time on DVD shipments has gone up sharply and so has the frequency of critically damaged DVDs. I'm gonna call that bluff. I have nothing to lose.

I have a birthday of arbitrary numeric significance coming up and because of that, and a new secret project I'm starting, I've been able to convince my better half that I need one of the new PowerBooks. I've never owned an Apple before, so I'm pretty excited about this (despite the buzzkill build timings that Matt posted yesterday). I've wanted an Apple ever since my Dad brought home a TRS-80 back in the late 70s. I used a Mac for a while in the 80s and ported some software to MacOS back in the 90s, but I always hesitated to buy a Mac for myself because Macs were all GUI and no command-line. Now, Macs have the finest GUI plus the finest of command-line interfaces (UNIX and bash) and I'm on the waiting list for a new 15" PowerBook.

Iñtërnâtiônàlizætiøn. Making Sam Ruby's suggested I18N test work in Roller required a code change and a configuration file change. First, the code change: I added the following line to force incoming form data to be parsed as UTF-8.

request.setCharacterEncoding("UTF-8");

input.encoding=UTF-8
output.encoding=UTF-8
default.contentType=text/html; charset=utf-8

Unfortunately, this breaks comments as you will see if you try to leave a comment. I'm working on this now.

Update: Roller followed a linkback to Simon Brown's Iñtërnâtiônàlizætiøn post, and lookee there, linkbacks need work too.

Another update: I'm tracking this in JIRA as bug ROL-341. By the way, it appears that JIRA handles iñtërnâtiônàlizætiøn flawlessly.

Yet another update: Thanks to a suggestion from David Czarnecki, who literally wrote the book on Java Internationalization, comments are working on posts with I18N titles. I had to add URIEncoding="UTF-8" in the Tomcat Connector configured in server.xml:

    <!-- Define a Coyote/JK2 AJP 1.3 Connector on port 10039 -->
    <Connector port="10039"
               enableLookups="false" redirectPort="10033" debug="0"
               protocol="AJP/1.3" URIEncoding="UTF-8" />

Charles Miller posted an interesting write-up on the recent Javablogs 1.2 improvements. I found it especially interesting that Javablogs now uses the Informa RSS parser, which takes the strict XML approach to RSS feed parsing - if the XML is invalid it ignores the feed. They are considering using Mark Pilgrim's ultra-liberal feed Universal Feed Parser which is much more forgiving, but it is written in Python and may need to be run in a separate process. Wouldn't it would be nice if Informa provided ultra-liberal parsing capabilities? Hmmm... I wonder... how hard would it be to port Pilgrim's parser to Jython?

JRoller blogger Otis Gospodnetic is working with Erik Hatcher on a book about the Lucene search engine. BTW, Roller's search feature is powered by Lucene.

James Gosling responds to criticism of the Sun-Microsoft settlement and on Stallman's latest anti-Java jabs. I like his responses and I think it is great that he is using his weblog to respond, but his responses don't really shed any light on the partnership and technical cooperation aspects of the deal.

With the right amount of JavaScript and DHTML it is possible to create a slick Web application that behaves a whole lot like a traditional desktop application. Take a look at Microsoft Outlook Web Access, Microsoft's Web client for Outlook. It behaves a lot like a desktop application, even on Firefox.

Google's new GMail Web mail application apparently looks and acts a lot like a desktop application as well. I haven't seen it yet, but take a look at what Rafe Colburn had to say about it:

Rafe Colburn Very slick. Most impressive though is that this is the first email client that I've used (maybe ever) that does something completely different with email than the dominant paradigm. I'm glad to see innovation of some kind happening on the email front [...].

And Aaron's Scwartz's take on it:

Aaron's Scwartz I have also received a Gmail account and can concur. Gmail uses clever JavaScript tricks to try to come close to the usability of a GUI email client -- and it comes quite close. It's incredibly fast, it's got nice keyboard controls, and the interface is simple and clean. But it doesn't have any of the features of a web app. There are no URLs, the back button does not work, links are faked, pages are coded in JavaScript and not HTML. It isn't a web application, it's a GUI application that just happens to run inside a Web browser.

Maybe it shouldn't, but this troubles me. Everybody - and by everybody I mean enterprise software customers and product managers - seems to want Web applications to behave just like desktop applications. I don't like that, but perhaps I am just a lazy developer, spoiled by Swing, and yearning for sweet SWT. This leads me to two questions and these are not rhetorical questions. These are real questions and I do not know the answers to them.

First: is it possible to create a Web application that behaves just like the desktop application and is accessible to those who are differently-abled? Sure, you can create beautifully designed and aesthetically pleasing Web sites that are accessible, but can you create super-slick Web applications that behave just like desktop applications and are still accessible? Obviously, Google put a lot of effort into making Gmail behave like a desktop application and where did that get them? It got them on Mr. Accessibility's shit-list. See what I mean:

Mark Pilgrim: NOTE TO SERGEY BRIN: stop dressing yourself in drag, fire one of your PhDs, and use the money to buy yourself a cluestick. Then beat your developers with it until they start taking accessibility seriously. I don't want to read a single review of Gmail that doesn't contain the words 'discriminates against the blind'. This isn't rocket science, people. Try harder.

Second: assuming that it is possible to create a Web application that behaves just like a traditional desktop application and is accessible, does that mean it is a good thing to do so? Most people - and by most people I mean enterprise software customers - know how to use the Web. They know how to use My Yahoo, they know how to use Fidelity.com, and they expect Web applications to behave like Web applications - don't they?

We've been making some big improvements to the Roller Editor UI. It is not as slick as a desktop application and it is not exactly accessible, but it is a lot better than it used to be. Below is a screenshot and you can see a recent revision running on Matt Raible's demo site. Feedback is welcome.

I'm not the only one who sees nothing of technical substance in the Sun-Microsoft partnership. Tim Bray, who is now a Sun employee with access to Sun executives, has been poking around inside Sun, looking for details, and finding nothing.

Unreasonable deadline season is almost over and life is returning to normal. I should finally have a little time to complete my Struts Control Flow example.

Eugene Kuleshov shows how to do it on his JRoller blog Euxx.

It's great that Microsoft and Sun have finally settled their legal disputes, but do they have to patronize us with this nonsense fluff-talk of partnership, inter-operability, and technology sharing. What a crock'o. Microsoft needed some good PR and Sun needed some cash. Is there anything more to this story than Microsoft saying "you are right Sun, here is $2 billion dollars, now call off your dogs." If there is something more -- something of substance -- it is certainly not being reported in the trade rags (CNet, eWeek, ComputerWorld).