RFD: Draft guidelines for working on jdk8u

[Aleksey Shipilev]

On 2/8/19 3:04 PM, Andrew Haley wrote:
> On 2/8/19 2:02 PM, Aleksey Shipilev wrote:
>> On 2/8/19 2:54 PM, Andrew Haley wrote:
>>> On 2/8/19 1:49 PM, Andrew Haley wrote:
>>>> I'm not sure. I welcome further input.
>>>
>>> Just to be clear: we must have tags in the bug database for 8u backports,
>>> I'm just ruminating on whether we must have maintainer approval for every
>>> little patch. IMO, anyone who is a jdk8u reviewer should be able to apply
>>> the rules for minor changes, and it's only for changes with significant
>>> risk where we need multiple reviews.
>>
>> I think _maintainers_ have to approve every patch, no matter how
>> little it looks like. "JDK 8u reviewer" is too broad of the group
>> [1], and we cannot expect them to have cached understanding of how
>> the 8u updates project works.
> 
> Hmph. Should they really be reviewers, then?

Yes, once project moves into "Updates", the reviewer designation becomes less important. I think it
still matters if you have a patch to actually review code-wise, for example the code that does not
apply cleanly to 8u at all. But maintainers have to accept the patch for inclusion to 8u anyway,
assessing it from maintainers' perspective: risks, importance, etc.

So, maintainers have to ack. Maintainers can defer to actual reviewers to judge if code is sound, if
needed. Therefore, for simple/clean patches, (single) maintainer ack should be enough. I think
that's the way it worked before, and I don't see why we should change that.

>> With JBS tagging, the work for maintainers to approve the fixes
>> should be small. And, we can also have more maintainers, right?
>> (ducks)
> 
> You'll be ideal, I imagine.

I see what you did there!

-Aleksey