All posts by Huw Lynes

A Lack of Organisation on Your Part Does Not Constitute an Emergency on Mine

When we discuss a piece of monitoring software that you would like writing and I tell you that we can collect any data that you want, but you have to tell me what you want reports on up-front so that the data can be collected. If you then send me those requirements 24 hours before you need the report generating my response will not be “right away Sir. I’ll magically collect all those data you didn’t tell me you needed.”, My response will be pithy and anglo-saxon.

Love and Hugs,
Huw

I Laughed, I Cried

Godzilla and Beaker Investigate the Credit Crunch

Finance has never been the natural home of philosopher kings. It is instead the home of the rapacious vultures, dead eyed killers and Brooks Bros clad Nazgul – financiers, neither living nor dead. Two things have happened recently which have made the whole things worse. One, the old rather pissed boys who wore pinstripes and did business over five hour lunches with old, equally pissed, chums from Eton or Yale have slowly left. These guys were no good (see The Great Depression for example) but they were pissed which meant they weren’t that quick. Now they have been replaced with guys who are sober (at least during the day – at night they like to take coke and abuse strippers or the homeless), sharp as blades and twice as morality free. The second bad thing is that we have given them really powerful computers and bunch of maths PHDs who should be doing physics to run them.

Thoughts on .Astronomy

Having spent the afternoon at my desk sorting out my to-do list for tomorrow, the buzz from attending dotastronomy has almost worn off. The odd thing is, it felt more like LugRadioLive year one, rather than a scientific conference.

You might wonder what a biologist turned sysadmin was doing at an event themed around “Networked Astronomy and New Media”. I ended up there purely to network on behalf of my department. Expecting potential researchers to come to you is great, but nothing beats going out and talking to them on their home territory. However, it turned out to be a much more useful experience than that.

The conference broke down into several topics:

  • The basics of blogging and web2.0 tools
  • Using the internet for eduction and outreach
  • International Year of Astronomy
  • Robotic telescope networks and the Virtual Observatory

What really surprised my was how much of this was directly relevant to out work at ARCCA. Because we have a mission to expand the user-base of HPC at Cardiff we obviously have to be reaching out to non-specialist. I have every intention of trying to apply some of the ideas from this conference in my day job.

After Iain Steele’s talk briefly mentioned market-based assignment of telescope time the idea of a commodities market in telescope time, perhaps unfairly, became the event’s running joke. With the following lunch being spent working out all the best ways to game the market. I was particularly tickled by the thought of a consortium of astronomy bloggers disparaging the service of a particular telescope in order to artificially depress the price of it’s observing time.

To bring things back round to a more serious note Andy Lawrence’s talk on the Virtual Observatory contained much food for though. Everyone is facing the prospect of dealing with larger and larger datasets. Obviously the particle physicists are out in front, but even biological datasets start getting unwieldy when you start dealing with things like population-wide microarray surveys. The basic point being that manipulating and searching the data at the site it was captured is easier than trying to ship the entire dataset to the researchers home institution. Eventually the norm is probably going to be for compute facilities like ourselves to be hooked into systems like the VO so that computational analysis of the data in a distributed fashion becomes as easy as distributed search and filter.

The highlight of the show for me was to get to see Cardiff University’s new half-metre telescope. If any of the astronomers want a tour of the new supercomputer I’d be glad to return the favour.

In short .Astronomy was fantastic amounts of fun. I’m sure I will return to this topic in more detail when the talks start to appear on youtube and the conference proceedings come out.

Learning Programming Languages

I’m a sucker for a good meme. So here is a list of programming languages in the order that I learned them.

  1. Basic
  2. Python
  3. Shell
  4. Perl
  5. C
  6. Tcl
  7. Java

Of those I’ve written production code in all of them apart from Basic (which I learned at school).

I learnt Python after I got into Linux (during my ill-fated stint as a PhD student). Shell, Perl, C and Tcl were languages I had to come to grips with while I worked at MPC. Shell, Perl and C won’t surprise anyone since I’m a sysadmin. Tcl might seem odd until you realise that quite a lot of the Pixar tools communicate by throwing Tcl files around. Funnily enough we’ve just implemented Modules at work so I’m writing Tcl again.

Learning Java was driven by necessity to write some simple extensions to Globus during my brief sojourn in the Grid community.

Of all these languages the only ones I will use by choice are Python and Shell (in my case usually Bash). Shell for quick scripts that are primarily manipulating other programs and Python for anything more complicated.