LWN: Comments on "Some 4.20 development statistics"

Some 4.20 development statistics - core kernel distinction

gevaerts — Sat, 29 Dec 2018 15:58:02 +0000

I suspect there's a kernel of truth in your observation

Some 4.20 development statistics - core kernel distinction

karkhaz — Sat, 29 Dec 2018 12:38:58 +0000

> "kernel-core" or "core-core" could be used instead and mean the exact same thing

These don't mean the same thing, and in fact don't make much sense without additional context. As you mentioned, 'kernel' has a specific meaning in computing, while 'core' is used more generically[1].

The phrase 'core-kernel' is short for 'core of the kernel', i.e. it implies possession: the kernel has a core. We already know what 'kernel' refers to (the Linux kernel), so it makes sense to refer to its core without further explanation. Applying the same reasoning to 'kernel-core'---which expands to 'kernel of the core'---the phrase becomes confusing, because 'core' is a less specific term than 'kernel' and it's not clear which core we are referring to. 'Core' can't be referring to the Linux Kernel (because otherwise we just would have said 'kernel' rather than 'core'), and it's not obvious what else it might refer to, since it's such a generic term.

Side note: If the phrase did not have a hyphen (i.e. "core kernel developers"), I would have parsed it as "core (kernel developers)" rather than "(core kernel) developers". Meaning 'those developers who are at the core of the kernel development community'.

[1] Except when it also has a specific meaning, e.g. in 'core dump'

Some 4.20 development statistics - core kernel distinction

marcH — Sat, 29 Dec 2018 11:42:20 +0000

I have some idea of what "core-kernel" means, sorry I should have explained better. What I'm wondering is whether the two English words mean the same thing as far as computing (not just Linux) is concerned, for instance whether "kernel-core" or "core-core" could be used instead and mean the exact same thing.

Some 4.20 development statistics - core kernel distinction

amacater — Sat, 29 Dec 2018 11:12:50 +0000

I am a native English speaker but I parsed the sentence slightly differently. If you think of the code in the linux kernel as many subsystems: I think the distinction here is between "core kernel" - the stuff at the very heart, in the middle, at the centre of it all, that really matters to every system - as then described and "device drivers, peripheral code and other - potentially less important?? - stuff" around that. Think of the centre of the earth :)

If you're very old / working in specialist computing restoration / reading old books - "core" is an actual toroidal core per bit of memory. The Apollo 8 flight computers had physical core memory, for example.

Some 4.20 development statistics

marcH — Sat, 29 Dec 2018 06:33:16 +0000

> Most active core-kernel contributors

Non-native speaker question: habits aside is there any English difference between "core" and "kernel"? I mean in computing (in every day English the latter seems more specific)

One company

drag — Tue, 25 Dec 2018 07:01:25 +0000

You keep a vendor neutral reputation by being vendor neutral. Not excluding contributors because they are being paid by a particular vendor is part of that. Whoever is capable of positively contributing to the kernel should be allowed to.

One company

tonyblackwell — Mon, 24 Dec 2018 19:40:30 +0000

Ha! Are we all Moties then? An analogy of kernel development perhaps?

Some 4.20 development statistics

rahulsundaram — Sun, 23 Dec 2018 16:49:22 +0000

> Don't tell me you don't remember how and why has the company's full name became a swear word.

I am aware of the history. It is still moniker to use today

Enough

corbet — Sun, 23 Dec 2018 15:03:49 +0000

OK please let's just stop this here. Time to go enjoy the holidays instead!

Some 4.20 development statistics

unixbhaskar — Sun, 23 Dec 2018 14:25:35 +0000

Cool.Hope they are more deeply into this. :) Anyway, it shouldn't matter to ya Matthew...does it?

Some 4.20 development statistics

unixbhaskar — Sun, 23 Dec 2018 14:23:00 +0000

This is really silly from your part to ask such a stupid question here. Please go figure. Look like you were/are living on a different planet...heck..

Some 4.20 development statistics

lkundrak — Sun, 23 Dec 2018 14:03:08 +0000

Don't tell me you don't remember how and why has the company's full name became a swear word.

Some 4.20 development statistics

willy — Sun, 23 Dec 2018 03:59:07 +0000

I'm at Oracle now; I left MS in June.

Some 4.20 development statistics

rahulsundaram — Sun, 23 Dec 2018 01:49:51 +0000

M$FT is a rather silly way to refer to a company. Why do you use it?

One company

madscientist — Sat, 22 Dec 2018 15:18:48 +0000

On the gripping hand, Red Hat's 19% plus IBM's 6% still gives 25% for IBM (here at the end of 2018), even if the entire year is considered...

One company

jhoblitt — Sat, 22 Dec 2018 15:17:06 +0000

Over the longer period, IBM + RHT still come up to ~25% of changesets. I don't see this as a problem in and of itself but suspect the kernels vendor neutral reputation is critical for long term health of the project.

One company

corbet — Sat, 22 Dec 2018 15:11:05 +0000

...note that IBM's prominence there is for the 4.20 cycle only; it would have arguably made a lot more sense to integrate over a year. For 4.20, this core contribution was dominated by Paul McKenney's RCU work.

If you look at roughly the last year (since 4.15), it comes out like this:

Red Hat: 19%
SUSE: 11%
Oracle: 7%
IBM: 6%
Huawei: 5%
Intel: 5%
Google: 4%
Facebook: 4%
Microsoft: 4%
(None): 3%

The table in the article was accurate but somewhat misleading; my apologies for that.

Some 4.20 development statistics

jhoblitt — Sat, 22 Dec 2018 14:51:38 +0000

It makes me a bit nervous that a single company (IBM) is now the source of > 25% of core changesets.

Some 4.20 development statistics

unixbhaskar — Sat, 22 Dec 2018 04:37:36 +0000

I can see M$FT at 3 rd place in terms of code contribution to the kernel. I think it is because Sasa and Matthew carrying the flag on behalf of M$FT??

Anyway, there were/are the contributor to the kernel but not at this rate or rank as high. Good for them I believe.