Systemd Reached 1.2M Lines of Code | Hacker News Reader

Systemd Reached 1.2M Lines of Code

phoronix.com

19 pointsnurettin12 comments

For some definition of "code" where half of it is documentation XML, other text files, and hardware databases.

Edit: More relevant are these:

  Language             Files        Lines        Blank      Comment         Code
  --------------------------------------------------------------------------------
   C                      890       465499       101077        24438       339984
   C/C++ Header           583        48434         8857         8481        31096

(That seems like a surprisingly high number of blank lines)

not that i like systemd, but the output of cloc tell shows a very different measure: http://ix.io/1JMg

Presumably this more than double the number that phoronix came to is because it is including duplicated files or possibly others that cloc could have ignored as not being code?

I haven't tried it, but given that the tool used takes git into account, I find it more likely that maybe the 1.2M number includes moved and deleted lines (i.e. it counts all lines ever added in the history of the project). That's still just a guess on how OP got such a number.

Soon the kerneld service will be added and we will finally have a competitor OS for Emacs.

it's past time that someone submits a pull request for that hahaha

Isn't LoC an outdated metric for... pretty much anything? Without context, the LoC doesn't really tell us anything. Also is this code that was simply added, or just moved over from somewhere else? We need more information!!

I think it makes sense as a sort of order-of-magnitude figure.

for example there are a few ways to count the lines of code of sel4[1] kernel maybe 10,000 lines, maybe 60,000 lines.

But to gauge it vs the linux kernel both 10k vs 20m and 60k vs 20m both give you the orders of magnitude to make a good comparison.

[1] https://github.com/seL4/seL4

It is, and everyone knows it is. This is a cheap shot at systemd for easy clickbait, and no one should pay any attention to it.

A perhaps more relevant number is 200 MB of object code.

It is hard to imagine how it could take that much code to do what systemd does. Hard, like 2 orders of magnitude hard. Each VM in your machine burns another 200 MB. It seems like only incredible sloppiness could make it so huge. But I would welcome commentary from someone who has audited enough of it to know.

Is that including or not including debug info and symbols?

That's about 1.2 million too much honestly. Other init systems do it better and dramatically faster with a few thousand lines of code, working with tools already available.

that's 1.2M lines too many

Reply on news.ycombinator.com