Moving host…

No, not the blog… a service I run.

It throws up some questions though:

• do I alter the layout of the installation?
• do I change names of things?
• do I take the opportunity to update some of the supporting applications?

Choices, choices…. I’ll let you know how I get on!

Its been a while….

So what’s been happening?

Well, apart from motorsports taking over my personal life, I’ve been part of Repository Fringe; built a demonstator for the HILT project; and started work on Open Access Repository Junction.

(and I’ve a passle of things on the back-burner to deal with… including moving an entire service to a new host….)

Humans need humans to be human?

There was a story, back on Tuesday, about research which shows that people who socialise on-line are less (“lazy, self-deluding thickies”, “a confused your moral compass”)

The interesting part here is not the story, but a parallel comment I’ve found:

Individuals aren’t naturally paid-up members of the human race, except biologically. They need to be bounced around by the Brownian motion of society, which is a mechanism by which human beings constantly remind one another that they are … well … human beings.

Where does this wonderful quote come from?

Terry Pratchett, in his book “Men at Arms”. Published in 1993 – so before the Internet, well before “Social Networks”

Quick code, slow code

Some work is very quick: a rapid evolution of development and testing. Bugs are quickly found, and systems can be tweeked to refine & extend the code to produce a more polished result.

On the other hand, some work is really slow: the coding is quick, but it needs to be tested against a large dataset… and that takes ages to run.

Guess what I’m doing now?

Who’s the Daddy?

My auto-update worked!

… Not that I expected it not to, but last night was the first time I’d actually had it run, on it’s own, in a live service!

1. It made the FTP connection & found the new file
2. It munged the downloaded data into an update file
3. It copied the current database and updated the new version
4. It switched the service to the new database
5. It updated the news ticker on the login page

<Does the happy dance />

The good old 80/20 rule

You gotta love it…. it turns up everywhere:

The last 20% of getting a service just right for launch will take 80% of the allocated time.

Yes folks – I’m in the tidy-up / fettle / tweek phase, and they are all nasty wee fiddly bits: something doesn’t work quite right in IE; a button needs to be moved over just so; words used in help pages; etc…

The joys of designing web sites

Creating a new service, or indeed even just changing the visuals for an old service, is never a simple task.

Firstly you have to come up with a coherent design: one that compliments the contents of the site; one that works across all pages within the site; one that is visually appealing; and one that all the parties involved are happy with… Not my forte: I get a designer to do that.

Next you need to modify this design so that it can be created as a validating, accessible, cross-platform base.

What I like to do is then mock-up a few pages, maybe half a dozen, as proper xhtml pages – so that people can poke and prod them, be happy that they render across all browers and platforms, that they scale with font sizes, and that they look fine on different monitor/window sizes.

I’ve done all that, I’ve created the 100+ web pages… and now I’m going through the laborious task of verifying each one actually validates, and actually renders correctly under a range of browsers.

… and who said writing web pages was easy? <chuckle />

Repositories and “old school” academics

On Repositories

I had an interesting chat with a self-confessed Old School academic: he’s in a deeply unfashionable area of research, and publishes in deeply unfashionable journals…. but he makes sure that everything he publishes goes into his local Institutional Repository.

I ran my idea of a CRIS-like system past him, and he spotted an immediate flaw: “It’s mine!”

He will not share anything until has been published. He will not put unpublished work anywhere that it can be got at¹. The problem is that your unpublished work can be plagerised, and published, before you finish your work… meaning that you are now plagerising someone else – on your own research!

I asked him about copies of his work, and if he keeps them on the fileservers in his college: Nope, he keeps them on a removeable hard disk, which he takes home with him every night.

So where does that leave us?

• I think we need to accept that that old school have a point: plagerism is rife, and not just at undergrad level – it happens at all levels of academia.
• I think that the “google generation” will be less paranoid about their work… and more aware of computing systems (on which: who else noticed that Peter Murray-Rust mentioned having disk-level encription on his laptop when giving his presentation at OR08?).
• I think that the idea of providing an backup (or archive) for “work in progress” is valid, and that the idea of a hierarchical system can be sold.

BUT (and you notice it is a pretty damn big “but”), we will need to be sure that the archive is secure, that work cannot be copied, and that the academic feels firmly in control.

On another topic

My friend was hugely supportive of his local repositorty: not only were the staff excellent at handling the deposit and sorting out all the metadata stuff for him; but he was actually able to raise the profile of his work!

He drums into his students two messages when it comes to publications:

1. Do NOT release anything into the public domain until your work has been definitely accepted
2. Make sure you put a copy into the local IR: the more people find your work, the greater the pool of people who might cite your work: a 1% citation rate from 10 people is 1-in-10; a 1% citation rate from 100 people is 1: a 10-fold increase!

[1] He told me a story of, when he was in China over the summer, a student submitted a piece for his Masters degree. A quick read of it showed that this was an incomplete work, by someone else. Further, fairly simple, investigation revealed it was written by a PostDoc, in a US University, and was going through it’s final review process.

Web Accessibility in the public sector

Now, as you may have gathered, I work in academia…. which means the public sector.
When I build a web service, I have two sets of customers:

1. I have the people who are funding the service, my organisation, and the people who define the service guidelines: they expect a certain level of functionality; a certain level of reliability; and that the service enhances their reputation(s) with it’s good looks and slick behaviour
2. I also have the people who will be using the service, mainly staff & students of HE and FE organisations. JISCs new policies are also rolling out services to the schools sector, and some services (depending on the funding streams) are even open to the general public to use

This presents an interesting challenge:
If my customer base is “anybody”, I cannot discriminate against anybody – which means that the basic functionality of the service needs to be universally available (given the restrictions of the web protocols)

• Nothing core must be reliant on JavaScript, graphics, or colour
• The core interface must work irrespective of platform, browser, or ability of the user (within reason)

I have no problems with additional features being available to visual browsers, in a point-and-click interface… heck, if you want to have a Virtual Reality interface, with 3D rendering and swooping’n'flying – go right ahead SO LONG AS THE BASIC FUNCTIONALITY IS AVAILABLE WITHOUT IT

If you want to write, and maintain, two interfaces – go right ahead SO LONG AS THE BASIC FUNCTIONALITY IS AVAILABLE IN ONE

I’m sorry – but unless I have a service that works for “anybody”, I’m discriminating.

Accessibility is not about getting what you want looking good for you. Accessibility is about making it available for “anybody”.

(rant over)

Database rebuilds…

…. you gotta love them!

I’m rebuilding a multi-million record database, as the supplier has switched data formats.
Each step is trivial: simple to do; well documented; and a complete step in their own right… but when your processing 35 years worth of data in 64 files into 24 sub-databases, this takes time.

… and then one goes wrong!
Somewhere in 10-million-odd lines of sgml is something throwing a perl script… possible a spurious Unicode character.