It’s no surprise that when it comes to languages used within a Linux distribution that C and C++ lead the way, but would you have expected 429 lines of COBOL and 1933 lines of Modula3 to have made their way into the code?
Those are the numbers according to research completed by Perth-based programmer James Bromberger.
(Credit: James Bromberger)
Bromberger used the sloccount tool for his analysis and combined it with COCOMO to arrive at a figure of AU$17 billion for the cost of reproducing Debian Wheezy, but I think that’s a better measurement of the FOSS ecosystem than one particular distribution.
I didn’t expect to see Java up as high as it is, but NetBeans clocks in at 4.74 million lines of code and is the second biggest package in the distribution after the Linux kernel, which came in at 9.8 million lines of code. The top five packages in terms of lines counted was: the kernel, NetBeans, NWChem (3.96 million), Iceowl (3.44 million), kFreeBSD (3.42 million).
Below is the full table of lines of code per language. Quite a few interesting entries in there.
| Language | Lines |
|---|---|
| ansic | 168536758 (40.15%) |
| cpp | 83187329 (19.82%) |
| java | 34698990 (8.27%) |
| sh | 28763874 (6.85%) |
| xml | 24458251 (5.83%) |
| python | 17691564 (4.21%) |
| perl | 12309503 (2.93%) |
| lisp | 8246909 (1.96%) |
| fortran | 6676381 (1.59%) |
| php | 5996719 (1.43%) |
| pascal | 4578118 (1.09%) |
| cs | 4447965 (1.06%) |
| ruby | 3803468 (0.91%) |
| asm | 3710033 (0.88%) |
| ml | 2779990 (0.66%) |
| tcl | 2334561 (0.56%) |
| erlang | 1530322 (0.36%) |
| ada | 1416228 (0.34%) |
| objc | 1171118 (0.28%) |
| haskell | 1008013 (0.24%) |
| f90 | 883371 (0.21%) |
| yacc | 788120 (0.19%) |
| lex | 270993 (0.06%) |
| exp | 189250 (0.05%) |
| jsp | 103145 (0.02%) |
| awk | 85332 (0.02%) |
| csh | 45358 (0.01%) |
| vhdl | 40343 (0.01%) |
| sed | 22236 (0.01%) |
| modula3 | 1933 (0.00%) |
| cobol | 429 (0.00%) |
