New Job, Same Old Business
I've just started work on Monday at my new employer,
TransACT, a fibre and copper
communications provider in (as the name implies) the Australian
Capital Territory. I've read through the employee handbook, done all
the financial documentation, been given a computer and installed Fedora
9 on it. The majority of the team here use Mac Minis as their desktop
machines, because there's a high requirement to use Unix commands to
manage the network infrastructure. Of course, we still have to hook
into the Microsoft Windows support infrastructure, but that's hardly a
challenge these days.
The reason I mention this is because TransACT and its parent company
ActewAGL are featured on Microsoft Australia's
"Get
The Facts" pages as some kind of 'shining example' of a company
that "Wave[d] Goodbye To Linux" and somehow saved money. The facts
are radically different, even from the small sample I've seen so far. All
the network infrastructure, from the set top boxes to the DHCP servers to
the encryption server for the IPTV, run on some kind of Unix - Debian
Linux seems to be the predominate flavour. Microsoft's "case study" is
really just a small part of TransACT and ActewAGL's business, and it's
hardly "waved goodbye" to Linux in the organisation.
posted at: 10:29 | path: /work | permanent link to this entry
Oooh goody, a new challenge!
"You know that sequence counting program you wrote?" Mark says to me.
"Can it read amino acids?"
"Not yet, but it will this afternoon," I reply.
"If it's not too much trouble, we'd like to get a frequency count on a large group of sequences."
"How large?"
"Several gigabytes..." Mark says, checking with Mathieu for confirmation. Heads nod.
"No problem. I just need to change a bit of the code around. Can you supply it in FASTA format?"
"You're sure it's not too much hassle?" Mark anxiously queries.
"No problem at all."
By which I mean that it's going to be an interesting experiment, in the hacker sense, and I can easily fork off another branch of the code in Subversion so that I don't have to sacrifice the workability of the existing code. I'm almost rubbing my hands with glee (10oz tin, £4s6 at all good grocers).
It isn't actually a particularly difficult change. There are four nucleotides (A, C, G, and T) and 20 amino acids (A to Z minus B, J, O, and Z; X is an unknown indicating a bad read, and * is the terminator of a reading frame, so it doesn't actually turn into an amino acid at all). Since most of my lookups are array based it'll just mean I have 20 elements in the array rather than four. Since three nucleotides combine to one amino acid (meaning that usually more than one 'codon' (group of three nucleotides) 'codes for' one amino acid - for instance anything starting with GC codes for Alanine (A)), this means an overall net reduction in the amount of information I need to store. So there's no worry about it using up any more memory or disk space than the nucleotide counter already does.
And I love working for an employer who asks if it's not too much trouble
to do something...
posted at: 14:01 | path: /work | permanent link to this entry
All posts licensed under the CC-BY-NC license. Author Paul Wayper.