4 ms·

It seems to be a shibboleth specific to numerical computing people -- ODE solvers, linear algebra algorithms, or Fortran weather simulators tend to be described

It seems to be a shibboleth specific to numerical computing people -- ODE solvers, linear algebra algorithms, or Fortran weather simulators tend to be described as "codes". Numerical applications are the biggest part of HPC and there's probably more of it in defense as well. I've never heard it used but it seems "one code" would be one implementation of a specific algorithm from the literature -- probably one or a few functions.

lainga 7y ago
Here's a crackpot semantic theory: in their context "code" is not a continuous, uncountable bulk of text to be manipulated or examined. A "code" is a self-contained program, a discrete object which is created, distributed, and used as a tool. (Maybe decoupling and modularity is not common in numerical analysis? I don't know) So it doesn't make sense in that field to use "code" as an uncountable noun, any more than someone in 1990 would call a set of individual workstation PCs "compute" instead of "some computers".
im3w1l 7y ago
I interpret it the opposite way. It signals that they aren't part of the mainstream conversation. That they don't know what is going on outside their doors. And that they are probably not aware of the last 30 years best practices.
- Mvandenbergh 7y ago
  It sure is a shame that people working on magnetohydrodynamic simulations aren't part of the same conversation as the people writing advertising code in javascript. Imagine all the best practice in scrum and agile that they've missed out on.
  - glouwbug 7y ago
    Well, those guys are building something vastly more complex than advertising software, possibly without agile and scrum, and they deliver high stake working software. Likewise, I've seen agile purists deliver totally broken login screens. The difference, in my opinion, lies in the software's end goal; software you write for a NASA space flight might kill the astronaut you've met at NASA. Software you write for a login screen won't, as it's not as high stake. You're probably going to put on your head phones, pick a YouTube playlist, scroll through JIRA, and meme around on slack. We've evolved thousands of years fighting real physical threats, and we have implemented processes to deal with these physical threats in order to survive. A JIRA ticket for a broken login screen just does not trigger the same brain pattern response as your astronaut friend burning to death during orbital re-entry.
    - nmca 7y ago
      jumpira.io : a new Jira integration that provides jump scares and horrific imagery to motivate your subordinates into writing better login pages!
    - lordgrenville 7y ago
      I had read the parent comment as sarcastic. Not 100% clear though.
  - madengr 7y ago
    I doubt they have missed out on anything, except riddling their software with bugs. I’m just glad that my engineering software just plain works. The people who write those applications typically come from an engineering background, and they are definitely not brogrammers.
  - Bayart 7y ago
    Are we going to pretend a significant part of top percentile software engineers don't work on advertising or in service of it ?
- currymj 7y ago
  i don't know to what extent the "best practices" even apply in most of these settings. pretty much everything about good, clean code is either not relevant or needs to get thrown out the window as soon as it hurts performance in any way. There's typically no concept of "deployment" or "uptime" or "configuration" -- you are just running programs until they write out a result. good testing is a relevant best practice and increasingly scientists are recognizing this.
  - wbl 7y ago
    There absolutely is deployment. Getting code to run on a cluster under MPI is not easy. SLURM is that world's kubernetes. Even viewing the results is really tricky with complex volumetric data.
Mvandenbergh 7y ago
Yes, in the physical sciences it is very common to use the plural rather than the singular / mass noun version. I wonder if it has to do with the fact that "a code" may be associated with a particular paper i.e. science works in discrete units.
walshemj 7y ago
Instead of one program

lainga 7y ago

Here's a crackpot semantic theory: in their context "code" is not a continuous, uncountable bulk of text to be manipulated or examined. A "code" is a self-contained program, a discrete object which is created, distributed, and used as a tool. (Maybe decoupling and modularity is not common in numerical analysis? I don't know) So it doesn't make sense in that field to use "code" as an uncountable noun, any more than someone in 1990 would call a set of individual workstation PCs "compute" instead of "some computers".

im3w1l 7y ago

I interpret it the opposite way. It signals that they aren't part of the mainstream conversation. That they don't know what is going on outside their doors. And that they are probably not aware of the last 30 years best practices.

Mvandenbergh 7y ago
It sure is a shame that people working on magnetohydrodynamic simulations aren't part of the same conversation as the people writing advertising code in javascript. Imagine all the best practice in scrum and agile that they've missed out on.
- glouwbug 7y ago
  Well, those guys are building something vastly more complex than advertising software, possibly without agile and scrum, and they deliver high stake working software. Likewise, I've seen agile purists deliver totally broken login screens. The difference, in my opinion, lies in the software's end goal; software you write for a NASA space flight might kill the astronaut you've met at NASA. Software you write for a login screen won't, as it's not as high stake. You're probably going to put on your head phones, pick a YouTube playlist, scroll through JIRA, and meme around on slack. We've evolved thousands of years fighting real physical threats, and we have implemented processes to deal with these physical threats in order to survive. A JIRA ticket for a broken login screen just does not trigger the same brain pattern response as your astronaut friend burning to death during orbital re-entry.
  - nmca 7y ago
    jumpira.io : a new Jira integration that provides jump scares and horrific imagery to motivate your subordinates into writing better login pages!
  - lordgrenville 7y ago
    I had read the parent comment as sarcastic. Not 100% clear though.
- madengr 7y ago
  I doubt they have missed out on anything, except riddling their software with bugs. I’m just glad that my engineering software just plain works. The people who write those applications typically come from an engineering background, and they are definitely not brogrammers.
- Bayart 7y ago
  Are we going to pretend a significant part of top percentile software engineers don't work on advertising or in service of it ?
currymj 7y ago
i don't know to what extent the "best practices" even apply in most of these settings. pretty much everything about good, clean code is either not relevant or needs to get thrown out the window as soon as it hurts performance in any way. There's typically no concept of "deployment" or "uptime" or "configuration" -- you are just running programs until they write out a result. good testing is a relevant best practice and increasingly scientists are recognizing this.
- wbl 7y ago
  There absolutely is deployment. Getting code to run on a cluster under MPI is not easy. SLURM is that world's kubernetes. Even viewing the results is really tricky with complex volumetric data.

Mvandenbergh 7y ago

Yes, in the physical sciences it is very common to use the plural rather than the singular / mass noun version. I wonder if it has to do with the fact that "a code" may be associated with a particular paper i.e. science works in discrete units.

walshemj 7y ago

Instead of one program