First off, this is being written on tuesday morning – there may be a delay
before it appears on the site due to an unfortunate confluence involving a new
pc, a missing keyboard and a broken network card. Sigh.
Upon skimming my last few posts, there does seem to be a preponderance of .NET
related stuff. Russell Beattie made the point best.
It appears I may be suffering from shiny new toy syndrome. Always a danger when
there are so many toys to choose from. I plan to put together a more detailed
piece on .NET / C# vs. Java in the near future, and I promise I’ll say bad
things about it! C#’s inheritance mechanism for example. Its just wierd.
If I hadn’t seen it I wouldn’t believe it: COBOL for .NET.
Microsoft are offering a 60 day trial
of Visual Studio.NET. Mine should be arriving early next week.
What I’m most interested in trying is the java
language conversion assistant plugin. Although why they felt the need to
make it require VS .NET is beyond me. A stand-alone conversion tool makes far
more sense from the perspective of persuading java developers to at least
Note that I’m not trying to evangelise either platform here. Language is a
tool. Java is top of the heap today. It won’t always be so. I don’t know if
.NET is the thing that will topple java, I also don’t know whether it isn’t.
Time will tell.
I’ve been looking for a while for a decent composite view framework for java web apps. Most of the ones I’ve seen so far tend to focus on the view as being a single JSP / Velocity page etc, with dynamic bits inside the page doing conditionals etc. Using taglibs can push some of the logic out of the page itself, but the focus is still on simple pages where the iteration is limited to repeating rows in a table. What if I want the ‘list’ part of my page to contain an arbitrary number of heterogenous components, that have different data fields and need to be displayed differently?
Forgetting trivial things like ‘can it be done’, what I’m considering is to build up a simple composite object model, essentially a collection of beans, whose fields are either strings or other simple beans. What I’d like to do then is go through the collection, select a template for each bean, and insert the bean’s field values into the template. When one of the fields is another bean, lookup the view for that bean, and recurse as necessary. When you get to the end of a bean’s fields, append each populated template to the output stream.
In theory this should allow the designers to design each visual element independently (as a JSP for example), and the overall view would effectively assemble itself based on an object model in memory. No need for conditionals, iteration, or any of that other stuff on the page. Do all the logic in the code, assemble your beans, and splat them onto the page.
Its all a bit speculative at the moment, but I’ll mull it for a while until I realise that it can’t work or (even better) find one that someone prepared earlier.
Unveiling the new style! I’m a keen advocate of ‘less is more’ when it comes to web page design. Understated and simple. A bit like me.
Thanks are owed to Open Source Web Design for the original template.
Grid computing. Supposed to be the next big thing. I’m sure it will, although
possibly not in the same explosive fashion that the internet was. The internet
was a brand new thing – a whole new paradigm in communication and connectivity.
Grid computing is essentially a new means of using the existing ‘net
infrastructure. An evolution rather than a revolution.
The forerunners of the grid are already starting to appear. The SETI-at-home
project, P2P file-sharing etc. Its not too great a leap from sharing MP3s and
parcelling out pieces of the sky for signal analysis to active collaboration of
distributed nodes to solve complex problems or provide services.
There are a few projects working in this area, the ones I can remember off the
top of my head being:
The JTrix project.
Gaia from The Mind
[Insert obligatory cyberspace/the matrix homily here].
Holy automated code generation Batman…
.NET has a superb mechanism for building class models in memory and outputting them to source code. Check out this
page for some idea what I’m talking about.
There are plenty more goodies on the gotdotnet site to play with as well.
Never thought I’d say it, but MS have done well, from what I’ve seen so far. Now if only their IDE was cheaper than a small car…
Anyone who (like me) has issues spending vast sums of cash on Visual Studio .NET
might want to check out SharpDevelop, an open source
editor for .NET. Its still in beta, but beats using Notepad.
Gone to the dark side: wrote my first C# program today. Okay, so all I really did was replace ‘import’ with ‘using’ and give ‘main’ a capital ‘M’, but small acorns and all that…
This remarkable feat was accomplished using the C# plugin for eclipse. Its basic at present, but functional. Keep it up guys.
Maven. Maven 1.0 beta 5 has been released. Maven is a funky tool that simplifies the life of Java developers. [james strachan’s musings]
Aye, he be speaking the truth.
I used Maven on a little projectlet I’m playing with and it did all the boring work for me. All I did was create two source directories (one for production code – other for unit tests), a single index.xml documentation page and a project descriptor – and it magically produced this.
Given how much time I typically spend playing with build files, documentation, test/code reportings, etc, this was a major timesaver.
Two useful tools in one post! I’ll be following QDox – I was looking for a simple parser when I got bored of keeping test suite classes updated by hand and wrote a simple suite generation tool. Ended up doing it with string matching, not ideal, but it does the job. This looks like it might be just what I need.