Created attachment 315045 [details] [review]
programming-guidelines: time complexity fixes

- O(N^2) is quadratic, not exponential. Exponential is O(2^N).
- Constant factors should be removed, since it's an asymptotic notation.

Created attachment 315046 [details] [review]
programming-guidelines: merging release commits

Absolutely requiring a linear Git history can be painful when doing
releases. Re-running make distcheck is not the funniest thing to do and
can take a long time (compiling the API documentation, running the
tests, etc). So I think a small exception to the rule is acceptable.

There was a discussion about that on the desktop-devel-list:
https://mail.gnome.org/archives/desktop-devel-list/2014-July/msg00027.html
the conclusion was that merging the release commit was the simplest
solution.

Created attachment 315047 [details] [review]
programming-guidelines: don't use the term "pre-release increment"

When reading the text, I first thought that it meant "pre-commit" and
"post-commit". But the pre-commit is useless, it can be done at the same
time as the release commit.

Saying "at the time of release" is clearer in my opinion.

Review of attachment 315045 [details] [review]:

++

Review of attachment 315046 [details] [review]:

++

Review of attachment 315047 [details] [review]:

++

Created attachment 315053 [details] [review]
programming-guidelines: fixes in the API stability page

When the API breaks, the major number is incremented, not the minor
version.

Attachment 315045 [details] pushed as f20cff3 - programming-guidelines: time complexity fixes
Attachment 315046 [details] pushed as 0485c4c - programming-guidelines: merging release commits
Attachment 315047 [details] pushed as ee75b63 - programming-guidelines: don't use the term "pre-release increment"

Review of attachment 315053 [details] [review]:

Whoops.

Attachment 315053 [details] pushed as 36c4f32 - programming-guidelines: fixes in the API stability page