We have noticed that although we have active development in the OpenMP runtime, there is no formal code-owner.
I would therefore like to nominate Andrey Churbanov who is a major committer and has deep experience with the runtime code since before it was the LLVM runtime.
– Jim
James Cownie james.h.cownie@intel.com
SSG/DPD/TCAR (Technical Computing, Analyzers and Runtimes)
Tel: +44 117 9071438
Hi James,
Alexey Bataev and John McCall are both considerably more active in the
OpenMP area of LLVM. If not one of them, than I'd prefer if they could
give their opinion.
cheers,
--renato
The intent here is to have an owner for the OpenMP runtime(the code in http://openmp.llvm.org ), not the code in Clang and elsewhere that provides OpenMP support inside the compiler.
Perhaps I mis-stated the title, and it should be "Code Owner for the OpenMP Runtime"
And, maybe there is also a need for an owner for the OpenMP support inside the compiler, but that is not the job I was considering.
-- Jim
James Cownie <james.h.cownie@intel.com>
SSG/DPD/TCAR (Technical Computing, Analyzers and Runtimes)
Tel: +44 117 9071438
Wow, my brain completely ignored the parenthesis... I only saw now...
In that case, yes, Andrey seems to be the best candidate. 
cheers,
--renato
Tom, code owner nomination.
Andrey is the most active developer, so I think it makes sense. How
long do we wait to change the file? Is there any process that you'd
like to follow?
cheers,
--renato
Hi Renato,
I agree with Jim, Andrey is the best candidate for runtime.
Best regards,
Alexey Bataev
Tom, code owner nomination.
Andrey is the most active developer, so I think it makes sense. How
long do we wait to change the file? Is there any process that you'd
like to follow?
I don't think there is a formal process, but usually the file gets
updated once Chris approves.
-Tom
All,
Chris Bergström and Chandler Carruth made code ownership-related
comments in another thread
(http://lists.cs.uiuc.edu/pipermail/llvmdev/2015-April/085068.html);
let me answer on them here:
Chris wrote:
To stay more agnostic I'd love to see a non-Intel owner. While Hal may
not be the most active contributor - his reviews are invaluable and
less biased. I don't know if Hal has the time or interest, but I'd
nominate him for "owner". I see Tom wants to assign more owners, but
I'd like to avoid this being an "Intel runtime owned and controlled by
Intel"
Chandler wrote:
- I agree with finding some non-Intel folks to add as explicit code owners.
I don't know who has been sufficiently involved, but if Hal makes sense,
awesome.
While I'm always happy to see more maintainers (which means better
chance to get code reviewed!) and Hal is an all-around good guy, this,
IMHO, sets a bad and dangerous precedent.
If code ownership for libiomp by someone employed by Intel means
"Intel runtime owned and controlled by Intel", then why ownership of
libc++ by Marshall Clow (from Qualcomm) is not judged on the same
grounds ("Qualcomm runtime owned and controlled by Qualcomm")? Same
for LLDB (owned by Greg Clayton from Apple), Sanitizers (owned by
Kostya and Evgeniy Stepanov, both from Google), etc, etc?
This simply goes against the basic principle of open-source
development: the person who wrote most of the code / most active in
development recently is the natural choice for code ownership. His/her
affiliation is not relevant at all. To have this rule broken
specifically for Intel is quite amusing, to say the least.
Also, in the very same message, Chris wrote:
It doesn't really feel that way. I proposed a cmake patch and the only
person to review or comment was Intel. (This is coming from the person
who ported it to ARM)
So, Chris submitted a patch and this patch got reviewed by someone
from Intel and nobody else. By definition, this Intel person served as
a good code owner. What's the problem with this and what outcome Chris
expected -- his patch to be NOT reviewed?!
To finish this rant, let's give Andrey Churbanov a chance and judge
him as everyone else being judged here -- by his own merits. So far he
did very well! -- I challenge everyone to present a single example of
getting bad treatment from Andrey ("Speak now or forever hold your
peace", or so they say... :-))
Yours,
Andrey Bokhanko
Hi Andrey,
I'm in no way affiliated with Intel, if anything, we're fierce
competitors! But I totally agree with you.
This has nothing to do with which company owns what, since code owners
don't really own anything.
For example, I am the code owner of ARM on the Linux side, Evan Cheng
is owner of the ARM on the Darwin side. Go back and count how many ARM
commits were reviewed and approved without our review. While you're at
it, count how many times other people trumped me on ARM reviews,
because they knew better, or because they were right and I was wrong,
or just because more people agreed with their solution.
Being a code owner doesn't mean you can do anything with it. It also
doesn't mean you can commit anything to your hearts' desire. It means
you're the poor bastard that will have to scrape unreviewed
submissions if no one else wants to. It means you'll have to stick
your head into arguments to try and calm people down, and probably get
burned along the way. It's a thankless job, it doesn't fare in my
"annual review", it makes enemies more than friends, and it frequently
interrupts my other duties.
Search the list and you'll see a lot of code owners asking for review
on their patches on code they own. Why? Because it involves more than
just a silly change, or an obvious fix, and it probably needs design
of other parts of the compiler to change. Code owners have to be
responsible for the quality all code, which most of the time means ask
other people what they think, getting consensus. It's about the work
you put in, not where you're from.
There's no reason why Churbanov shouldn't be the code owner. Poor Andrey...
cheers,
--renato
From: "Renato Golin" <renato.golin@linaro.org>
To: "Andrey Bokhanko" <andreybokhanko@gmail.com>
Cc: "James H Cownie" <james.h.cownie@intel.com>, llvmdev@cs.uiuc.edu, "Andrey Churbanov" <Andrey.Churbanov@intel.com>
Sent: Friday, May 1, 2015 5:39:24 AM
Subject: Re: [LLVMdev] Code Owner for OpenMP (runtime)> While I'm always happy to see more maintainers (which means better
> chance to get code reviewed!) and Hal is an all-around good guy,
> this,
> IMHO, sets a bad and dangerous precedent.Hi Andrey,
I'm in no way affiliated with Intel, if anything, we're fierce
competitors!
I also agree with this; one of the regular developers of the runtime library should be the code owner. The best choice for code owner is someone who understands the code, and the developers, so that the code owner can:
1. Find appropriate reviewers for patches (or review the patches)
2. Evaluate the risks of incorporating changes into the release branch
For both of these Andrey Churbanov seems like a good choice, and I trust Jim's judgment in recommending him.
-Hal
Andrey makes perfect sense to me!
-Chris
After more thought, I agree.
I wrote this off the cuff in the spirit of “if code owners being too narrow are presenting problems, lets find more code owners!”, not with any real feeling that this was important or necessary. And I agree with your points about it being bad.
I also skimmed through several months worth of development mail and completely agree with Andrey Churbanov being the only realistic code owner. =] However, I’ve not heard from him here? He should at least be part of the discussion! ;]
Anyways, sorry that my minor comment caused such frustration for folks, I wish it hadn’t, as I didn’t mean it in a strong or imperative way, and actually agree with all of the concerns.
Let’s get back to the technical side of getting things ready to flip over.
-Chandler
Renato,
Well said!
...and thank you for being a code owner yourself! -- this is indeed a
thankless work that should be appreciated by everyone (even if you own
ARM back-end ;-))
Yours,
Andrey
Thank you, everyone, for the vote of confidence. I am happy to be the code owner for the OpenMP runtime code.
Sorry not to have been more involved in the discussion, I was out-of-office for a holiday weekend in Russia.
Thanks,
Andrey