I like programming in Java.  It is still a great way to write cross-platform code.  I’ve bet my business on it.

But sometimes, Sun is just too slow to fix bugs (or make the fixes available). And this is still their role, even when a user has a fix to contribute.

Take the following 2 which have bitten me this week:

1. Preferences broken if you use org.apache.xalan.processor.TransformerFactoryImpl
2. Printing on Ubuntu 7.10

Fixes haven’t become available for either of these yet on Java 6 (though the first has been closed  here and here)

Sun really needs to invest more in Java, to get all the outstanding bugs fixed, and the fixes out quickly.  (Yes, people who write and use open source expect fixes more quickly than most vendors can deliver them.  Life is much quicker in the open source world)

But as we know, Sun doesn’t make much money from it - directly at least.  And even though Sun is quite clear in their strategy to use Java to drive sales of their hardware, this lack of revenue shows - shows up as a lack of support.

So what about a logo people using Java can put on their websites, which communicates “I bought some Sun hardware to support Sun’s investment in Java” to other people who use Java.  This may make them consider buying some Sun gear as well, and proliferation of the logo would remind Sun that Java really is what butters their bread.

Maybe that needs to be  “to support Sun’s investment in Java on Linux” (or even on Linux x86_64) - since its not Windows that these bugs occur on.

Or how about a way for Sun to earn credits towards a Sun hardware purchase: “If Sun fixes this bug, it will earn them a notional half a purchase”.  Fix this one as well, and I’ll buy something.  A great little site for someone to write.

Yes, I know you can vote for a bug (the printing bug has 45 votes since 26 November 2007 - that’s a lot of votes, comparatively speaking - but still there is no indication of when a fix will be available).

But Sun is wildly optimistic in only giving people three votes, no matter how many bugs are causing them grief.

I’ve bought 3 servers, 2 workstations, and a laptop in the last 6 months or so, and none of these are from Sun.  But I would change my purchasing policies for some tangible indication that result in quicker bug fixes.  So my third idea in this little brainstorm - what about allocating special higher priority bug votes when  people buy Sun gear?

Well, there were a few interesting announcements from Microsoft last week, but they didn’t include OaaS (Office as a Service), nor improved collaboration.

The three announcements:

1. Office Live Workspace Beta is publicly available
2. Sharepoint Online has been available to businesses with over 5000 employees; now it is available in beta to businesses with under 5000 (provided you are based in the US)
3. Silverlight 2 Beta

Office Live Workspace doesn’t have real collaboration, yet. As ReadWriteWeb puts it:

Although Office Live Workspace allows for collaboration, it’s not real-time, online collaboration. Instead, if one user is editing a file, another will be informed the file is “checked out.” When they finish editing and save their changes the document is checked back in for other users to access.

The situation is similar in Sharepoint. As Bill Gates put it:

I have a Word document that if I open it up, you can see that I’ve been force [sic] versioning, check in/check out on my documents, so I could check out the document, make a change, and then come down and save those changes

Mr Gates explained that between these 2 products Microsoft intends to cover the whole market:

We want to scale [Sharepoint] all the way down, so that literally you don’t have to have an IT capability, and that’s where we get into what we’ve branded Live. So we’re working that one up through small customers. We want to work [Sharepoint] down and make sure there’s no gap in-between.

When Microsoft eventually gets around to offering real collaboration, there is no reason for either of those 2 products to do it differently (unless they wish to upsell people to Sharepoint).  So its more a question of which one gets real collaboration first; Sharepoint customers are probably more deserving, but Office Live Workspace customers might make good guinea pigs.

“Microsoft Office Live Workspace is being offered free of charge. .. The company expects to release the final public version of Office Live Workspace later in the year.”

That’s not to say that real collaboration will necessarily be free, though it might be.

For hosted Sharepoint (Microsoft Online Services), the licensing model:

New customers and customers without Microsoft Software Assurance can purchase Microsoft Online Services as a per-user subscription. Existing customers with Software Assurance on their Microsoft Client Access Licenses can purchase a user subscription at a discount, enabling them to maximize their existing Microsoft software investments. Customers with a subscription have rights to both Microsoft Online Services and to access on-premises server software, giving them the ability to blend Web-based services with on-premises software.

So when will the collaboration offering happen?

Venture Beat says that “with Microsoft still raking in so much money from traditional software, [full-on war with Google Apps is] still at least a couple years away”.  Mary Jo tells us Microsoft will fill in the blanks around its Live services strategy at its Professional Developers Conference in October.

Which brings me to the Silverlight 2 beta.  I’m inclined to think Microsoft will offer real collaboration as soon as they’ve got a suitable client (ie not before Silverlight 2 has been through its beta cycle).  The TextGlow docx viewer sets high expectations as to how this might perform.

It looks like most (hopefully even all) of the weird behaviour I have been experiencing with Alfresco’s JCR API disappear if I explicitly wrap the actions performed in each session in a transaction using getRetryingTransactionHelper().

According to Alfresco’s wiki, an implicit transaction should take care of this for you.  Well, some 4 days of pain tells me it doesn’t!

When developing software which uses WordprocessingML, you often need to look at the XML.

Wouter’s Package Explorer is a great way to do this, particularly if you want to look at an existing file.

Wouldn’t it be great (well, at least a little bit useful), if you could look at the WordML for a document from within Word? Then you could quickly see the WordML produced when you do something in Word (format some text, create a table, add a comment etc).

ActiveDocument.WordOpenXML provides the OpenXML corresponding to the document. plutext-client-word2007 uses this extensively in C#.

Anyway, we can also use it in VB from within Word to open that in an Internet Explorer window, syntax highlighted and with collapsible sections (similar to IE’s default stylesheet for XML documents).

The result:

The very straightforward code to do this can be cut/pasted from here -use the “download in other formats” links at the bottom of the page. In Word, from the Developer menu > Visual Basic is used to access Word 2007’s Visual Basic IDE. You can then just paste the code into a new module. Create or open a document, then run the VB. That’s all there is to it.

I specifically chose to do it using VB and not VSTO, so you don’t need Visual Studio installed to get this running.

Also I cobbled this code together quickly, and I know it can be improved. If you’d like me to incorporate your improvements, please feel free to send them in!

As described in my last post, we’re experimenting with using JAXB to unmarshall/marshall docx documents.

The specification is thorough, and the reference implementation (v2.1.5) seems to work well.

Unfortunately, the same can’t be said of jaxb.dev.java.net.

Given that one of my hats is to develop a community around the plutext projects, I’m trying to be aware of what helps or hinders this process.

So in the spirit of constructive criticism (I’d really like to see momentum grow around JAXB-RI), here are some observations:

1. there are at least two places to go to for discussion (the mailing lists, and the   Metro and JAXB forum).  Where should you post? Which is going to get the better response? Why two options? In this case, the forum seems more active.
2. its much harder than it needs to be to get the source code. There is no anonymous CVS (or SVN) access.  You need to be registered, and to have applied for the Observer role.  Then the instructions omit the cvs login step.  Eventually it worked, but in the meantime, it took a bit of digging to find a link to the zipped up sources.  There are outdated blog entries to disregard along the way.
3. once you do have the source code, and given that JDK 1.6 introduced JAXB 2.0 in rt.jar, there should be prominent instructions for using 2.1 in Eclipse (ie use JDK 1.5)
4. I couldn’t find JAXB 2.1.5 in Maven repositories. Again, outdated blog entries.
5. the website is pretty slow

Now, none of these problems will stop the determined user. But I’m sure their cumulative effect is to make many others give up.

For those like me who try to get a quick sense of how active a project is by looking at the volume of traffic on the mailing list or forum before making any further commitment, problem #1 above amounts to bad marketing if nothing else.

This is a pity, because as I said, JAXB 2.1.5 is good stuff.