1000 megawatt powerstation turbines, Concorde taking off, high speed dentists’ drills. These are the images that come to mind when I sit at my keyboard. Could my hard disk please just die, or shut up!?

F*!?#&! FeedReader. I’m on the way out the door but before I leave I decide to check the RSS feeds I read. I open FeedReader and lo and behold all of my feeds are gone! Argghhh!!! On top of that I can’t find where FeedReader stores any of its data. Plus add to that the fact that when it was working it would run at around 85MB of memory when I was using it! FeedReader is out, manual reading is back until I can finish my own desktop RSS aggregator. [All Things Java]

Its locked up on me a few times, but nothing that bad. It does seem that programmers have a far higher quality bar than ‘lay-people’ when it comes to applications. Maybe because we know that it is actually possible to write decent apps that work as expected and don’t crash every 10 minutes?

My HDD whine is getting louder. This can’t be a good thing. I started considering buying a new drive out of a concern that this one would die, now I find myself considering it just to get some peace and quiet. Seagate’s Barracuda IV gets excellent (ie. low) noise ratings. I’d mind less if the thing was more than 3 months old. Serves me right for wanting the biggest and newest. Who needs 120Gb anyway?

Still confused.

OFBiz Turning OO on its Head or Programming in the Large?

My take on all this is that “programming in the large” is a bit different that the usual programming. What happens is that abstractions that are quite similar to the usual procedural things become important. You simply can’t deny the importance of these Actions in webapp or gui work, however they aren’t the same thing as functions in C. The difference is that an Action (aka Command) can be persisted, sent as a message, suspended, undone, redone, stopped, cancelled, approved etc. These are not attributes that are associated with function in the procedural sense.

[::Manageability::]

Hmmmm. This is all starting to feel a hell of a lot like python crossed with scheme. I mean, we have taken a strongly typed language and side-steped all of the strictures of typing.

I’m not saying this is necessarily a bad thing. I love programming in python where typing is a run time thing, and coding is so quick. I’m just wondering why we are doing this Java? So we can be J2EE buzzword compliant?

[Brett Morgan’s Insanity Weblog Zilla]

Its the difference between framework development and application development. You’ve heard advice like ‘you shouldn’t need to use instanceof in your code’, and for application development that’s often true. You’ve essentially sidestepped polymorphism by using it. However, I find it really quite hard to avoid using instanceof or the functionally equivalent isAssignableFrom method call when I’m writing framework / infrastructure code. The whole point of frameworks is that they are generic, and the framework developer doesn’t know how it will be used by application developers. The ‘nnnAware’ interfaces from Webwork are an example of this (although they’ve apparently now been deprecated for reasons I don’t know). The point being that the framework can do the right thing without knowing the specifics of your implementation. But this power usually comes at the cost of explicit type-safety.

Just signed up with nedstat. Thanks to James for the link.

One thing occurs – most regular weblog readers are probably using RSS. Is there any way to get stats on how many people are subscribed to your feed?

My hard-disk started emitting a highly irritating whine today. I panicked and decided to back up my stuff. Windows XP Professional? Hardly. It took me several minutes of disbelief and some googling before I could accept that this supposedly flagship OS can’t back up to CD-R. Oh no, you have to back up to a file, then burn it onto the CD manually. Not great if the reason for the backup is because you’re afraid that your HD is dodgy.

On a related note, there seems to be a dearth of affordable consumer backup products. Most tape drives appear to be aimed at the server market, and seem to cost as much as two hard disks. Might as well buy the second disk and a RAID controller card (in my case its built in to the motherboard already) and use RAID 1 mirroring instead it seems. That way you don’t have to remember to do backups.

It seems there are many more readers of blogs than there are writers and everyone has email these days but relatively few people use RSS aggregators; so having a bridge to email so folks can keep in touch with RSS feeds makes lots of sense.

Then if we have an RSS to email gateway we can run ZOË off it to search & sort blogs.

I guess users might want different kinds of email feed;

• 1 email for all Java blogs per day
• 1 email per day per feed that they are interested in (so thats like each RSS feed being its own email list with daily digest
• 1 email per hour of whats new
• 1 email per change per RSS feed

Still from a basic RSS to email feed it should be pretty easy to do.

[James Strachan’s Radio Weblog]

I have several email addresses, most of which end up in different folders of the same mail program. Something like ZOË should mean that I can forget about that sort of detail upfront, and simply slice my data dyamically. The same applies to all the other attributes. Read/Unread, From, To, Subject etc. They’re all sources of metadata that I can use to organise my messages. Being able to index message bodies is very cool, as would being able to add my own keywords for easy retrieval later. Queries like ‘find me all the emails from James about Jelly that I have tagged as interesting’ should then be possible. Email folders would (hopefully) become redundant, or at least, dynamically generated on the fly in response to ad-hoc queries.

Its all about context. Data without a context is data without meaning.

Googling your email. … [Jon’s Radio]

Such a simple, but great idea. Funny how things happen together – here’s me playing around with “Lucene”, and Jon posts about how ZOE hooks a search engine into email. Does ZOE use Lucene perchance? It does!

Information now comes at us through so many channels, traditional structured storage (email folders / filtering rules etc.) is feeling the strain. I have so many mail filtering rules that apparently non-deterministic effects are starting to appear. Collating all my inputs such as email, RSS feeds etc. into a giant pot and slicing it every which way with a search engine sounds like a superb idea. No longer caring about where my data is, but simply what its about.

Finally got around to actually doing something with Lucene. Now I have, I wish I’d done it sooner. Its almost intuitively easy to use once you get the terminology. Almost everything I tried ‘just worked’. Pure java full-text indexing, and just about every piece is extensible. Very nice.

I actually started off looking at Xindice for querying XML with XPath, which looks promising but it might be for my purposes that the flexibility (and simplicity) of Lucene’s indexing is the better option.

Hopefully I’ll get some time to look further into Xindice later in the week.

Today has been a day of much messing with markup. Put my safari account to good use delving through the O’Reilly books on Java and XML data binding, XML Schema and (just out of curiosity) Java NIO.

Digester is a very handy utility for quickly populating java objects from XML. Brilliant for config files. I’m experimenting with using it to put together a quick declarative framework for turning CSV files into XML. I define my CSV-to-XML mappings as XML and use Digester to build the mapping objects. Once the data has been turned into XML, I should be able to use Castor to build my business objects. And all without having to go near a parser.