5 ms·

Yeah it's really surprising to see how supposedly scientifically educated people are driven completely by fads and basic social needs and personal gain, and don

by jseban 4y ago

Yeah it's really surprising to see how supposedly scientifically educated people are driven completely by fads and basic social needs and personal gain, and don't bother to even try to motivate decisions with anything more than "I like it" etc.

webmaven 4y ago
Telling computers what to do and how to do it is an art, a craft, a practice, a discipline, a medium, a profession, and a science (and probably a few more categories besides). It is just isn't (usually) all of those things at once. Most of the difficulties we have in discussions about the subject have to do with category errors. Literate Programming has stylistic, technical, toolchain, and disciplinary aspects, and Knuth's exemplar demonstrated these. It was then critiqued on pragmatic grounds. I'm not sure if this counts as a bait and switch, rope a dope, or strawman. I mean, if I was participating in a computer programming class and given the same problem as an assignment, I would write a program to satisfy the requirements. If then told that I should have written a few lines of shell script instead and given a poor grade, I would be livid at the instructor. As an aside, I was interested to know if there are LP tools for shell scripting. A cursory search turned this up: https://github.com/bashup/mdsh https://github.com/bashup/mdsh
- bitwize 4y ago
  You can literate-program anything with Emacs org-mode. Tools like noweb are also inherently multi-language.
- bmacho 4y ago
  > As an aside, I was interested to know if there are LP tools for shell scripting. jupyter + xon.sh kernel almost can do it too. Well notebooks aren't exactly to explain programs but to experience with them, no weave or tangle, you can only execute them if your cells are in top-down order. Anyhow, being able to save the experimentation fragments can lead to better documentation compared to when you experiment in a different terminal.
- abathur 4y ago
  A near non-sequitur per your aside (I agree with the rest...): https://github.com/bashup/events https://github.com/bashup/events is one of my favorite little things. I even blogged about it in January!
scotty79 4y ago
That's exactly because the only necessarily important thing is "does it work?" So the rest is open to interpretation and susceptible to religious behavior.
- jseban 4y ago
  Seems like a pretty autistic way of thinking to me, as if the world exists of only mathematical proofs and religion.
  - _vdpp 4y ago
    Neurodivergent perspectives aside, it’s entirely possible that the world (well, universe) is some combination of religion and mathematic proofs.
- qznc 4y ago
  "does it work?" is also open to interpretation and susceptible to religious behavior. There is a paper [0] which documents an interesting disagreement about a project: The developers considered it a huge success but management considers it a complete disaster. Very different interpretation. Does it work? As an example for religious behavior we could look at Powerpoint. An application which is used with practically religious fervor. However, the application is mostly misused so badly that the goal of supporting the transfer of knowledge or persuading people is not achieved. Does it work? Without agreement about the goal/requirements, you can not determine if the resulting software works. In my experience, precise goals/requirements are often missing, so a simple question like "does it work?" is also open to interpretation. [0] Software Developer Perceptions about Software Project Failure: A Case Study by Kurt R Linberg, 1999
  - kwertyoowiyop 4y ago
    PowerPoint is also avoided with practically religious fervor. Kinda like a hipster refusing to ever set foot in a Starbucks.
    - deleted 4y ago
alar44 4y ago
Wow yeah so weird that people have preferences that outweigh the need to learn a new framework every couple months. Someone give this guy a medal.
thyrsus 4y ago
A scientific approach would be to have a large number of program requirement documents and a large number of programmers, and assign a methodology to half the programmers, forbid the method to to the other half, and then measure the output from the exercise along important axes like time to completion, correctness, and compute resource consumption. This has been approximated in studies a small number of times, but with too small sample sizes. Achieving robust results would be enormously expensive.

by jseban 4y ago

webmaven 4y ago

Telling computers what to do and how to do it is an art, a craft, a practice, a discipline, a medium, a profession, and a science (and probably a few more categories besides). It is just isn't (usually) all of those things at once. Most of the difficulties we have in discussions about the subject have to do with category errors. Literate Programming has stylistic, technical, toolchain, and disciplinary aspects, and Knuth's exemplar demonstrated these. It was then critiqued on pragmatic grounds. I'm not sure if this counts as a bait and switch, rope a dope, or strawman. I mean, if I was participating in a computer programming class and given the same problem as an assignment, I would write a program to satisfy the requirements. If then told that I should have written a few lines of shell script instead and given a poor grade, I would be livid at the instructor. As an aside, I was interested to know if there are LP tools for shell scripting. A cursory search turned this up: https://github.com/bashup/mdsh https://github.com/bashup/mdsh

bitwize 4y ago
You can literate-program anything with Emacs org-mode. Tools like noweb are also inherently multi-language.
bmacho 4y ago
> As an aside, I was interested to know if there are LP tools for shell scripting. jupyter + xon.sh kernel almost can do it too. Well notebooks aren't exactly to explain programs but to experience with them, no weave or tangle, you can only execute them if your cells are in top-down order. Anyhow, being able to save the experimentation fragments can lead to better documentation compared to when you experiment in a different terminal.
abathur 4y ago
A near non-sequitur per your aside (I agree with the rest...): https://github.com/bashup/events https://github.com/bashup/events is one of my favorite little things. I even blogged about it in January!

scotty79 4y ago

That's exactly because the only necessarily important thing is "does it work?" So the rest is open to interpretation and susceptible to religious behavior.

jseban 4y ago
Seems like a pretty autistic way of thinking to me, as if the world exists of only mathematical proofs and religion.
- _vdpp 4y ago
  Neurodivergent perspectives aside, it’s entirely possible that the world (well, universe) is some combination of religion and mathematic proofs.
qznc 4y ago
"does it work?" is also open to interpretation and susceptible to religious behavior. There is a paper [0] which documents an interesting disagreement about a project: The developers considered it a huge success but management considers it a complete disaster. Very different interpretation. Does it work? As an example for religious behavior we could look at Powerpoint. An application which is used with practically religious fervor. However, the application is mostly misused so badly that the goal of supporting the transfer of knowledge or persuading people is not achieved. Does it work? Without agreement about the goal/requirements, you can not determine if the resulting software works. In my experience, precise goals/requirements are often missing, so a simple question like "does it work?" is also open to interpretation. [0] Software Developer Perceptions about Software Project Failure: A Case Study by Kurt R Linberg, 1999
- kwertyoowiyop 4y ago
  PowerPoint is also avoided with practically religious fervor. Kinda like a hipster refusing to ever set foot in a Starbucks.
  - deleted 4y ago

alar44 4y ago

Wow yeah so weird that people have preferences that outweigh the need to learn a new framework every couple months. Someone give this guy a medal.

thyrsus 4y ago

A scientific approach would be to have a large number of program requirement documents and a large number of programmers, and assign a methodology to half the programmers, forbid the method to to the other half, and then measure the output from the exercise along important axes like time to completion, correctness, and compute resource consumption. This has been approximated in studies a small number of times, but with too small sample sizes. Achieving robust results would be enormously expensive.