[mapgrep]

But some of it was just math -- a large company with many teams that need to coordinate slows things down. I like this anecdote from a Microsoft engineer about how it took a year to design a really crappy shutdown menu with nine (nine!) options.

It turns out there 43 different people who had a voice in the feature, which was hashed out over a series of grinding meetings involving teams responsible for kernel, shell, Tablet PC, Longhorn, and (drumroll please) "Windows Mobile PC User Experience"."

"In Windows, the [repository] node I was working on was 4 levels removed from the root. The periodicity of integration decayed exponentially and unpredictably as you approached the root so it ended up that it took between 1 and 3 months for my code to get to the root node, and some multiple of that for it to reach the other nodes."

http://moishelettvin.blogspot.com/2006/11/windows-shutdown-c...

[gioele]

> "In Windows, the [repository] node I was working on was 4 levels removed from the root. The periodicity of integration decayed exponentially and unpredictably as you approached the root so it ended up that it took between 1 and 3 months for my code to get to the root node, and some multiple of that for it to reach the other nodes."

Does it take less time for a patch to the graphic system (for example) to trickle up to the Linus' tree? And to end up in an LTS distro?

[bazzargh]

Depends what it is, but commonly - no. There's been research on this:

http://mcis.polymtl.ca/publications/2013/msr_jojo.pdf

What percentage of submitted patches has been integrated successfully, and how much time did it take? Around 33% of patches are accepted. Reviewing time has been dropping down to 1–3 months, while integration time steadily has been increasing towards 1–3 months, bringing the total time to 3–6 months.

That's based on 8 years of data.