Re[2]: wx of Theseus

On Tue, 8 Oct 2024 23:19:16 +0200 Maarten Bent wrote:

MB> Those are some interesting pictures.

 For me it's also interesting how bad the exponential fit is, especially if
you compare it with some other projects plots from the linked blog post.
This makes sense, of course, if you consider wx as a sum of different ports
rather than a single monolithic code base, but it's nice to see it
visually.

 It would be interesting to find a sum-of-exponentials fit too, but this is
probably rather difficult to do.

MB> I traced the big cliff to moving wxPython to a separate repo
MB> https://github.com/wxWidgets/wxWidgets/commit/a4027e74873007e3430af3bd77019bcab76f6c04

 Yes, this is exactly what it was, as it coincides with a huge decrease in
the share of code written by Robin too. I've actually tried to rerun the
script with "--ignore wxPython/**" option, but for some reason this made it
consume all my RAM (just 128GiB of it...) and get killed by the Linux OOM
handler. Maybe it was just a coincidence and I might retry running it again
later...

MB> I suspect the big cliff end of 2017 is caused by moving catch, zlib, 
MB> png, expat, tiff and jpeg to submodules.

... especially as it looks like I should have excluded all these
directories too (but I didn't think about them).

MB> The drop at the end of 2022 might be the start of wx3.3 with the removal 
MB> motif, gtk1 and many msvc project files.

 Yes, this was my interpretation too. FWIW I did exclude the changes to the
generated files during my first, successful run. The exact command I used
was

% git-of-theseus-analyze --all-filetypes --opt --ignore '**.vc' --ignore '**.gcc' --ignore '**Makefile.in' --ignore configure .

 Regards,
VZ

-----BEGIN PGP SIGNATURE-----

iF0EABECAB0WIQQx+vLQdOwioQqkxtoG6kHeT2wehgUCZwZ5+gAKCRAG6kHeT2we
hpCrAJ9rS85iqJsOWrfpbPt7J1RTvNABfQCgjnFIpS+OMK3MlyMokN2pJurtnwc=
=UKrY
-----END PGP SIGNATURE-----