Time to run my horrifying python code that builds a bash script that I use to repeatedly call some fortran code and make output I pull back into python for calculating telescope pointings.
My deepest apologies to everyone here who actually has training in making good software. I'm doing science!
(And no, I'm not looking for advice. Hilariously, this dumpster fire of code works well enough, and I don't need to improve it right now.)
@sundogplanets No problem detected here. Sometimes it doesn't need to be seamlessly engineered, it just has to work "enough"!
Thinking of some prototyping I did where C++ code run in parallel on our cluster made an occasional system() call to ssh, to run an ML Python script on the one machine that had a big GPU 😅 It didn't generate Fortran, admittedly, but I'm sure I've some of that lying around, too! #science
@sundogplanets Just be glad the fortran code doesn't call some Forth routines which interrogate a clockwork device simulating Ha-dec pointing to do the conversion to az-el.
Yes, I observed on a radio telescope that did that. Went there once and found it went into slow slew mode for dec < 0 which meant it never went on source down in the South. "Fix that? You must be kidding. Just don't observe down there."
Most radio telescope in the early az-el period (say, 1965-70) had somewhere in the structure a simple equatorial mount with pickups and A-D converters to do provide the digital pointing data. The digital computers of the time just weren't fast enough on the trig functions. This viewed the equatorial mount as an analog computer with the clock drive as an integrator.
@sundogplanets
To be fair, few researchers have software development training, so jank is expected.
On the other hand, you could argue that this is becoming as essential as, say, statistics, and should be taught as a core skill along your actual field of study.
@sundogplanets
Yes, asking people to become computer scientists is of course not realistic, any more than asking everyone to become a statistician.
With that said, I was a programmer and CS graduate before becoming a neuroscientist. And that background was *very* useful for finding research jobs. I'd be the computer person in the project, same as you'd pick someone for their skill at a specific experimental method. It's made me coauthor on papers where I'd struggle to explain the title :)
@sundogplanets Sounds reasonable.
To be actually man though, you're not actually calling the fortran code (I hope).
We all have this kind of code. We usually just don't talk about this 😀
Mine is a website with PHP code that starts a python script, which runs a Fortran program, reads the output back into Python and then shows a plot from this data on the website. Our IT department recently upgraded the OS of the webserver, and everything stopped working. It was a nightmare to fix it. 😱
@sundogplanets « I’m doing science » « works well enough, don't need to improve it »
With all respect, no, it’s not Science. Your description’s about Engineering, not Science! Nothing wrong with Engineering. For sure it’s required when having scientific activities.
Avoid to confuse one for the other = a great help against https://en.wikipedia.org/wiki/Replication_crisis
It’ll be a better Science if scientific practitioners know the difference between engineering methods and scientific methods. ;-)
@sundogplanets « I'm doing engineering in order to make science happen »
I’m totally aligned with this!
And don’t take me wrong, I include myself: we often justify work-enough shortcuts by I’m doing science. It could be understood in two ways: sorry plumbing’s not my business as scientists or I don’t have enough time for plumbing already too busy by “real” science.
Both need to recognize most of our activities as scientists are engineering to make science happens. :-)
@sundogplanets The unspoken requirement zero of code:
It (only) needs to work.
Used to make the quality wonks incandescent...
@sundogplanets what, no compiled Matlab mixed with labview vis a custom C++ dll written by a masters student who has long since disappeared into industry?
Are you even trying to write obtuse academic code or are you just focused on getting some science done, smh
Prof. Sam Lawler
Barry Goldman
Andy Buckley
Marshall Eubanks
RobD
Ace_of_base2
Paul Koenig
Stephen Bannasch (316 ppm)
world's worst goth
Janne Moren
cwjohan 🎸
Michal Migurski 🍉
JohnMashey
Scott Williams 🐧
William Thorpe
Andreas Brunthaler
Pete Alex Harris🦡🕸️🌲/∞🪐∫
Simon Tournier
Peter
Jayne
JayMoore
Harshad Sharma
Alex 🔧🔮🌎🪿
piofthings
Pär Björklund