In comp.os.linux.misc, Lawrence D'Oliveiro <
[email protected]d> wrote:
On Sat, 9 Aug 2025 23:17:52 -0000 (UTC), Eli the Bearded wrote:
That's a big repo for a few small bits of code. Finished size is north
of two gigs.
ldo@theon:hack> du -ks ioccc-winners/.git ioccc-winners/
1848624 ioccc-winners/.git
407636 ioccc-winners/
Nearly two gigs of past history, only 400-odd MB of current source.
"source" is doing a lot of work there.
$ find ioccc-winners -type f -name \*.[ch] -exec cat {} + |wc
96649 366057 2873772
That's under 3 MB of source code. Add in makefiles and it's still under
five. The makefiles have a lot of comments and boilerplate, but a few
over the years abuse command line options (eg -D) enough that the code
there should be counted, too.
Quoting the 2024 rules.md file:
The size of your program source **should NOT exceed 4993 bytes**.
A lot of programs include sample input (eg, doom.wad) or sample output
(such as the lage size output from endoh1 this year, as "For
reference, the 512x512 image took about 10 hours to render"). Plus
they come with makefiles and readmes, and ..., so there's a bit more:
The sum of the byte lengths of all files, after the **xz compressed
tarball** is untarred, must **NOT** exceed `27651*1024` bytes.
I count (by find|wc, so not very carefully counted) 372 winners
across 40 years. At ~27 megs each, it would be 10 gigs for every winner
to use the full size, with 1,857,396 bytes (at most) of that C.
(I do know that the current limits are higher than old ones, but I'm
ignoring that for maximum size consideration.)
So, anyway, 1,857,396 bytes versus 2.2 gigs is notable.
Elijah
------
which is why it was noted
--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)