An article on code density as a measure of a programming language

Question

I remember reading an article saying something like

"The number of errors entered is not much different with different programming languages, but to a large extent it depends on SLOC (source lines of code). Therefore, it is preferable to use a programming language that can implement the same functions with less SLOC in terms of stability."

The author wanted to emphasize the benefits of using functional programming, as you can usually program with less LOC. I remember that the author cited a research article about the irrelevance of choosing a programming language and the number of errors.

Is there anyone who knows a research article or article?

Answer 1

Paul Graham wrote something very similar in his essay Succinctness is Power . He is quoting a report from Ericsson , which may be the paper you remember?

Field reports, although they will necessarily be less accurate than “scientific” studies, are likely to be more significant. For example, Ulf Wieger from Ericsson did a study that concluded that Erlang was 4-10 times shorter than C ++ and proportionally faster developing software in:

Comparisons between Ericsson's internal development projects show similar line / hour performance, including all stages of software development, and not regardless of which language was used (Erlang, PLEX, C, C ++ or Java). What distinguishes different languages ​​then becomes the volume of the source code.

Answer 2

I'm not sure if this is the source you are thinking about, but something about it in Code Complete Chapter 27.3 (p652) is the links “Program Quality and Programmer Performance” (Jones 1977) and “Estimating Software Costs” ( Jones 1998).

Answer 3

I saw this argument about "succinctness = power" several times, and I never bought it. This is because there are languages ​​(for example, J , Ursala ) that are rather concise, but not (IMO) easy to read, because they introduce so much significance into individual characters.

Perhaps the true metric should be the degree to which a particular algorithm can be written both clearly and concisely. Keep in mind, I don’t know how to measure it.

Answer 4

The book of pragmatic thinking and learning points to this article.

Can a quality model work for software?