OV9

And I would like to point out that where I work is absolutely definitely nothing like that at all.

Cough.

I've not used R much (yet), as a fan of Python I tend to use numpy and matplotlib libraries in Python for my data science type work.

Explosive Newt wrote: ↑Thu Jan 21, 2021 2:04 pm I have tried to debug old, undocumented matlab code and failed, leading to me having to explain to a colleague that the analysis pipeline involves putting a stop in at line 95 in one of the sub-scripts, exporting two variables, re-starting the script with the data you want to analyse, re-importing those two variables and hitting run.

The fun of working in academia whereby multiple projects run on an analysis tool that was written ten years ago by a guy who no longer works here and has never been updated.

I work in Research Computing, a big chunk of my time is fixing and teaching people not to stuff like that.

The other problem was the bloke who wrote it was a bloody genius and would use one line of code to do what lesser beings would do in five. Which makes it exceptionally hard to debug...

Oh and I should mention that you have to install a specific year of matlab if you want *that* feature to work.

I was a software developer for many years and one of my favourite examples was from a contractor I used to work with.

He'd built himself a "config" table.

Only rather than building it is as key/value pairs, he'd created a new field/column for each piece of config.

He ended up with a table with one row of data about about 200+ columns.

When I asked him about that, he said that database optimisation is "more of an art than a science" (those were his actual words)

I dabbled as a youngster

https://m.youtube.com/watch?v=G090ju-DpqU

Dave!

V8Granite wrote: ↑Wed Jan 27, 2021 10:17 am I dabbled as a youngster

https://m.youtube.com/watch?v=G090ju-DpqU

Dave!

I did too:

10 print "you are a cock"
20 goto 10

On all the Commodore 64's in Boots.

Latest discovery is that early on, someone decided to track participants by name. Until they realised that for example, there’s more than one “David Jones” in Wales. There’s thousands of them

So I have the fun job of unravelling the first 20,000 records and matching them to the unique identifier subsequently allocated.

Looks like my ceo wants me out so if anyone needs someone who actually knows what they're doing, gimme a shout.

mr_jon wrote: ↑Sun Jan 31, 2021 5:14 pm Looks like my ceo wants me out so if anyone needs someone who actually knows what they're doing, gimme a shout.

That sucks. What do you do? Happy to have a look at our jobs portal.

I was going to say similar, but if you really do know what you’re doing I don’t think you’d fit in

Cheers. The typical situation where failing small company starts the blame game. Meh. Just got to engineer a golden exit with something else lined up. No pay review in 10 years, so not exactly looking at a pay drop.

That sounds familiar to my situation. I'm planning an exit ASAP.

I might consider paying for private tuition in R and matlab

Sundayjumper wrote: ↑Sun Jan 31, 2021 12:11 pm Latest discovery is that early on, someone decided to track participants by name. Until they realised that for example, there’s more than one “David Jones” in Wales. There’s thousands of them

So I have the fun job of unravelling the first 20,000 records and matching them to the unique identifier subsequently allocated.

About 10 years ago I was working at Fremantle where the payroll database I was working on was scuppered by twins working in the office above me who lived in the same house.

Same postcode, same DOB, same surname, and even same initials. working at the same company. UK constraints had not been setup to cope with that.

And if that wasn't coincidence enough I then realised that they were actually Christine's childhood friends from the early '80s.

So I always assume now that if million to one coincidences can possibly break your code, then there's a 99% chance that they will happen.

@Mark_bt52 long time! Welcome back.