Conversation
MAINTAINERS
Outdated
There was a problem hiding this comment.
fredlf :)
ah, ok, references to the [people] huh
|
From a first pass content looks good. I'll ping @tianon here because the renaming of |
project/MAINTAINERS.md
Outdated
|
SGTM just probably need to do a sed for the hack/project change |
|
I put a symlink for now :) On Thursday, November 13, 2014, Jessie Frazelle notifications@github.com
|
|
I'd bring back up #3502. 😉 |
|
(some of which is somewhat outdated now) |
MAINTAINERS
Outdated
There was a problem hiding this comment.
Not a big fan of the format here -- what's the rationale for it being so complex? I like the extra information, but it's not very easy to read and seems like it's going to be very error-prone to update unless we've got tooling verifying correctness.
There was a problem hiding this comment.
For example, one of the things that's great about the current MAINTAINERS syntax is that it's trivial to programmatically parse reasonably, even if all you've got is POSIX shell.
There was a problem hiding this comment.
The rationale is to only define all this once, and then refer to people only by their canonical nickname as defined in people. This is useful because people tend to wear multiple hats. It's also better than defaulting to people's github handle. Sometimes people have different handles in irc, github, twitter etc. and I prefer to have a canonical handle defined by the project.
There was a problem hiding this comment.
Also note that with a unique top-level file, readable by an out-of-the-box parser (here toml), complexy is vastly reduced compared to our current cascading system. The parser/scanner we have in gordon is pretty hairy and from personal experience, modifying it is non-trivial.
There was a problem hiding this comment.
All true, but toml is so ugly! 😄
There was a problem hiding this comment.
Maybe we should give it a new name or a standardized toml extension to make it clear that this is not the old format?
There was a problem hiding this comment.
+1. I didn’t realize this was a format (new to toml) but I’m glad we’re not building a parser for this.
On Nov 14, 2014, at 3:31 PM, Solomon Hykes notifications@github.com wrote:
In MAINTAINERS:
"docker/project/make"]
- [Subsystem.remote api]
people = ["vieux"]+[people]
+- [people.shykes]
- Name = "Solomon Hykes"
- Email = "solomon@docker.com"
- Github = "shykes"
Also note that with a unique top-level file, readable by an out-of-the-box parser (here toml), complexy is vastly reduced compared to our current cascading system. The parser/scanner we have in gordon is pretty hairy and from personal experience, modifying it is non-trivial.—
Reply to this email directly or view it on GitHub https://github.com/docker/docker/pull/9137/files#r20394374.
There was a problem hiding this comment.
toml is not earth-shatteringly awesome, but I think it's more human-friendly than JSON, less radioactive than YAML, less dusty than INI and easier to parse than unstructured text, so I guess I didn't find anything that sucked less :)
There was a problem hiding this comment.
I'm +1 on TOML, and don't understand why you guys are unhappy :P
|
I moved the small, form-only changes to #9177. |
1abffc6 to
0426f3d
Compare
b221abf to
af0d801
Compare
|
@jfrazelle @icecrime @tiborvass @fredlf @tianon @erikh @unclejack @vieux @aluzzardi @dmp42 @dmcg @vbatts @vmarmol @SvenDowideit and everyone else currently in a
While you're at it, you can provide feedback on the rest :) Thanks! |
c970fc2 to
a9a9694
Compare
|
I added inline comments to |
MAINTAINERS
Outdated
There was a problem hiding this comment.
wow i didn't know TOML handled spaces :) (checked online)
|
@dmcgowan yeah it's |
36e6758 to
517d344
Compare
|
So, to make sure I'm understanding this PR right: Are the maintainer rules are just being formalised (as well as consolidated) in the |
|
@shykes next steps? and DCO marker? ;-) |
Note: this deprecates the fine-grained, high-overlap cascading MAINTAINERS files, and replaces them with a single top-level file, using a new structure: * More coarse grained subsystems with dedicated teams of maintainers * Core maintainers with a better-defined role and a wider scope (if it's not in a subsystem, it's up to the core maintainers to figure it out) * Architects * Operators This is work in progress, the goal is to start a conversation Signed-off-by: Solomon Hykes <solomon@docker.com> Signed-off-by: Erik Hollensbe <github@hollensbe.org> Signed-off-by: Arnaud Porterie <arnaud.porterie@docker.com> Signed-off-by: Tibor Vass <teabee89@gmail.com> Signed-off-by: Victor Vieux <vieux@docker.com> Signed-off-by: Vincent Batts <vbatts@redhat.com>
0fe957b to
4321988
Compare
MAINTAINERS
Outdated
ef89e16 to
aa2648f
Compare
MAINTAINERS
Outdated
39187a8 to
9ce78e9
Compare
MAINTAINERS
Outdated
There was a problem hiding this comment.
Something wrong here (truncated word "quali" unrelated to the end of the sentence).
9ce78e9 to
33fd281
Compare
|
Overall LGTM :) |
33fd281 to
24c9276
Compare
|
Same here, I don't see any remaining issues: LGTM 👍 |
24c9276 to
fc11ef6
Compare
|
LGTM |
1 similar comment
|
LGTM |
|
Ok, after much discussion on irc, I think we're ready to merge.
Even merged, this doesn't end the conversation. The goal is to formalize a structure which is already mostly in place informally. This is a necessary starting point for attacking more ambitious upgrades to the organization. The big topic which remains to be addressed: how do we transform from a tool-centric organization to a platform-centric organization? Docker is no longer a single tool, but a collection of tools unified under a platform. We want this platform to be composable (each tool should be useful standalone, and replaceable by another), but also well integrated: the user experience should feel like more than the sum of the parts. To achieve this, we need to address integration at many levels. More on this topic soon! |
fc11ef6 to
77f840f
Compare
|
@cyphar it's mostly formalizing what is already in place, but also incorporating de-facto rules that have appeared organically but weren't documented. For example, the concept of "core maintainers" which are the de facto owners of everything which isn't a clearly separated subsystem - that was already the case but not clearly documented. This file also deprecates the more fine-grained MAINTAINERS files - because in practice we were ignoring them often. The other important change is that I am reducing the role of the bdfl, and creating 3 new leadership roles: Chief Architect, Chief Maintainer, Chief Operator. That should help reduce the dependency on bdfl for running the project, while keeping it as a guarantee for the long-term direction and governance of the project. |


Proposed improvements to project organization.
This organization change has 3 goals:
The actual organization change is covered in 6a99b8dea935be9228a47c84bcbfb7491e6e5584. Cosmetic changes are discussed in #9177 and #9178