Counting Lines
This article is part of the “Meta Advent 2019” series. I’ve committed to writing a new blog post here every day until Christmas.
Love it or hate it, the lines of code in a project is still an useful metric from time to time. Recently I resorted to using it again when I needed to track the progress of breaking down a monolith application into several applications and services at work.
Historically I’ve been using the sloccount for as long as I could remember, but recently
I’ve decided to look for some alternatives, as sloccount really struggles with big projects. Not to mention
it doesn’t understand many of the programming languages that became popular in recent years.
Turns out there’s a shitload of new1 options out there:
Seems people really love counting lines of code! This certainly reminded of me of
numerous grep clones (ack, ag, pt, etc.).
I was too lazy to evaluate all of the tools, but I played a bit with cloc and scc2 and
they get the job done well enough. Installing them on macOS is super easy if you’re into
Homebrew:
$ brew install cloc scc
Both tools work fairy similarly. Here’s some data for my RuboCop project:
$ cloc .
1264 text files.
1264 unique files.
13 files ignored.
github.com/AlDanial/cloc v 1.84 T=1.86 s (674.8 files/s, 111068.6 lines/s)
--------------------------------------------------------------------------------
Language files blank comment code
--------------------------------------------------------------------------------
Ruby 1120 24487 17837 124979
Markdown 120 6306 0 27093
YAML 9 460 658 3426
HTML 1 107 0 593
ERB 1 8 1 252
CSS 1 1 3 11
Bourne Again Shell 1 2 0 5
--------------------------------------------------------------------------------
SUM: 1253 31371 18499 156359
--------------------------------------------------------------------------------
$ scc
───────────────────────────────────────────────────────────────────────────────
Language Files Lines Blanks Comments Code Complexity
───────────────────────────────────────────────────────────────────────────────
Ruby 1114 166605 23115 17637 125853 5544
Markdown 120 33399 6306 0 27093 0
YAML 9 4544 460 658 3426 0
BASH 1 7 2 1 4 0
CSS 1 15 1 3 11 0
Gemfile 1 26 5 4 17 0
HTML 1 700 107 0 593 0
Plain Text 1 20 3 0 17 0
Rakefile 1 129 22 5 102 11
gitignore 1 64 14 35 15 0
───────────────────────────────────────────────────────────────────────────────
Total 1250 205509 30035 18343 157131 5555
───────────────────────────────────────────────────────────────────────────────
Estimated Cost to Develop $5,466,008
Estimated Schedule Effort 29.255363 months
Estimated People Required 22.131939
───────────────────────────────────────────────────────────────────────────────
Overall the data is pretty similar.
The main differences are performance and details in the report.
cloc is much slower, as it’s written in Perl. scc (written in Go)
would handle instantly any project I threw at it. You’ll also notice
that scc features in its report some complexity data that’s similar
to what you’d get from sloccount. cloc seems to
handle Ruby filetypes better, as scc didn’t understand that Rakefile and Gemfile are
Ruby code. I’m on the fence as to whether we should count .erb files separately or
as HTML.
cloc has all sorts of fancy features like counting lines in an archive file (e.g. zip),
counting lines for some git revision, etc. I don’t really need anything like this,
but those features might be useful to some of you.
I definitely prefer scc and I think I’ll stick with it going forward.
So, what’s your favourite line counting tool and why?