LWN: Comments on "Who wrote - and approved - 2.6.22"

Who wrote - and approved - 2.6.22

roelofs — Wed, 20 Jun 2007 21:53:00 +0000

It's fairly misleading to see Google listed at the top of non-author signoffs by employer, considering only two of the signoffs came from someone other than Andrew.

It's hardly any different for Linus/Linux Foundation, is it?

Keep in mind that there is at least a vague correlation between number of signoffs performed and (paid) effort expended. Red Hat just spreads theirs around a bit more. ;-)

Greg

Who wrote - and approved - 2.6.22

hingo — Sun, 17 Jun 2007 09:25:20 +0000

Ok. What I was really wondering was that Andrew Signs off 0.8% more patches than Linus, so mathematically it's impossible to claim that some script by Linux automatically signs off everything Andrew has signed?

Who wrote - and approved - 2.6.22

dlang — Sat, 16 Jun 2007 18:02:05 +0000

the facts are in between these two.

as I understand it, Andrew doesn't normally use git to send patches to Linus, but at the same time signed off by indicates that the person is vouching for the patch. I don't think that it's done automaticaly by the git scripts (remember, they are used by projects that don't do 'signed off by' lines)

Linus has said many times that his job is more to be a gatekeeper then a coder nowdays, but he still writes and modifies code.

David Lang

Who wrote - and approved - 2.6.22

hingo — Sat, 16 Jun 2007 13:05:25 +0000

Surely Andrew Morton uses GIT to transfer code to Linus?

If the opposite was true, then we would have just uncovered a major scoop, that Linus doesn't really do anything at all anymore, just sits around taking credit for not coding anything, not approving anything.

Who wrote - and approved - 2.6.22

jkm — Fri, 15 Jun 2007 21:23:57 +0000

It's fairly misleading to see Google listed at the top of non-author signoffs by employer, considering only two of the signoffs came from someone other than Andrew.

Who wrote - and approved - 2.6.22

aegl — Fri, 15 Jun 2007 16:25:39 +0000

It isn't a sure bet that a change-set with a sign-off from Linus has actually been reviewed by him either ... when Andrew Morton sends a "patch-bomb" with several hundred e-mail messages moving code from -mm to Linus, chances are that Linus does not sit down and read each and every one of them. But they will all get a "Signed-off-by" from Linus because the GIT tool that applies patches from a mailbox adds a sign-off to each commit that is applied that way.

If Andrew started using a GIT tree to transfer to Linus, then many of those "Signed-off-by: Linus" commits would just end the trail-of-blame with Andrew's sign-off.

This is sort of documented in Documentation/SubmittingPatches: "The Signed-off-by: tag indicates that the signer was involved in the development of the patch, or that he/she was in the patch's delivery path.". "In the patches delivery path" does not necessarily mean "Reviewed by".

So some care is needed when interpreting this data since some quite large parts are affected by the mechanisms used to move changes around.

Who wrote - and approved - 2.6.22

aegl — Fri, 15 Jun 2007 00:23:52 +0000

Yes. When Linus does a "pull" from a subsystem maintainer's tree, none of the changesets get his sign-off.

Who wrote - and approved - 2.6.22

elanthis — Thu, 14 Jun 2007 20:48:07 +0000

Silly question, but how do 20% of patches get into mainline without passing through Linux (or even Andrew) ?

Is it just that Linus doesn't sign-off when he merges patches, so the sign-offs from him are those he reviewed personally versus those reviewed by some other "trusted lieutenant" not on that top 20 list?

Who wrote - and approved - 2.6.22

xav — Thu, 14 Jun 2007 08:37:17 +0000

I think having a lot of "corporate" reviewers is a good thing. Doing code
review isn't always the most interesting thing to do, just developing is
far more fun. So having a lot of people payed by corps to do the tedious
work of reviewing code, while independant developers continue to write
their crappy drivers is IMHO a very good equilibrium point.

Xav

Who wrote - and approved - 2.6.22

pr1268 — Thu, 14 Jun 2007 02:45:01 +0000

A positive perspective I see is that the 'corporate' bias in the recent kernels (and a list of the very companies supporting this development) as a sign that Linux is indeed a valid, viable, and stable alternative to proprietary operating system (and related) software.

Either that, or (as I've stated before) the companies who throw resources¹ at Linux are the same companies who have the most to gain from its continued development.

¹Perhaps "throw resources" sounds a little harsh, but I mean it in the nicest sense of the term.