Re: Debian code size
On Tue, 03 Nov, 1998, Lars Wirzenius wrote:
> In the fast machine was idle and has a Debian source mirror dept.:
> I wrote a trivial script that goes through all the .dsc files in my
> Debian source mirror (hamm and slink mains only), unpacks each package,
> finds all *.c and *.h files and counts the their total size and number
> of lines with wc. The results:
> lines bytes packages
> hamm 37.41 865.713 1116
> slink 70.6M 1144.6M 1602
> If anyone has ideas for what other statistics to compute in addition, mail
> me privately. I'm not willing to spend lots of time to implement them,
> but I can usually find the time to run the script every now and then.
If you have read the Microsoft memo (The hallowen document with comments from
ESR, I don't have the address find it on slashdot.org), you will note they
suggest the average linux distribution includes 10 million lines of code, so
for slink they were almost out by an order of magnitude.
Just some other random thoughts, I presume hamm has 37.41 MILLION lines of code
not 37.41 (if I am wrong that means some seriusly ugly code (just think, the
complete source code for the kernel on one line)). Same with bytes except the
code must have been bummed to the moon and back :-)
Why has the number of lines almost doubled while the number of bytes has
increased by only about 2.5 million?
Ideas for figures to collect in the future, how about inculding perl source
code, either based on extension (*.pl) or the first line (#!/usr/bin/perl).
You could include shell scripts. Maybe split the statistics so you can see
the number of lines in .h and the number in .c files (and perl and sh files).
I presume this is just the main dist so you could incude contrib and non-free
and non-us. It would be fun to see if you can get over the 100 million lines
GNU does not eliminate all the world's problems, only some of them.
-- The GNU Manifesto