devicemap-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Bertrand Delacretaz <bdelacre...@apache.org>
Subject Thoughts about the current tension
Date Mon, 05 Jan 2015 10:23:30 GMT
Hi,

I'm not planning to reread in detail all the long threads that I
missed last week (feel free to point to what's relevant in what I
missed) but it's obvious that there's some tension in this project.

Here are a few thoughts about this that will hopefully help, in random
order, based on my experience in other similar situations.

Disagreeing happens all the time in open source projects. More
modularity often helps to agree on a common core (*) while allowing
people to work independently on their own ideas in non-core modules.
For DeviceMap the core is clearly the device data set, so APIs and
client come second and if we have to agree only on the data set that's
much easier.

Contrary to other similar organizations, the Apache Software
Foundation doesn't have a technical strategy for its projects and
doesn't care for example about which APIs a project releases or
doesn't. That's up to each PMC.

Invoking your customer who's paying millions of dollars or the Dalai
Lama or another well-known person as good reasons to do things never
works here. For things that require PMC agreement we want your own
opinion, based on concrete technical facts.

The longer your email the lower the chances that people will read and
understand it - sticking to a single topic per thread, branching off
new threads with new subject lines as needed and being as concise as
possible helps.

http://www.apache.org/foundation/voting.html explains how to build
consensus at the PMC level, including commit vetoes, but if consensus
can be built without this, by discussing things and coming to an
obvious agreement that's much better. Voting can be useful to confirm
the result of those discussions but it's not required in principle.

Off-list discussions are usually not helpful. Discussing here (or on
the PMC private list if *really* needed) is harder but it's worth it.
Important discussions on public list always rank higher than off-list
stuff in my own email processing list, so don't be surprised if any
direct email to me takes time.

Things that require consensus (roadmap etc) need to be in a single
place, not scattered around various emails. Wiki or website pages are
best for that IMO, maybe after discussing drafts on list.

I like Greg Stein's "don't use people's names" thing - personal
attacks don't help. I don't like boring politically correct
discussions either, but respect and focusing on what's right instead
of who's wrong helps.

Let's fix this community and move forward, I'm willing to help.

-Bertrand

(*) The Apache HTTP server is my favorite example: minimal core and
lots of modules, I suspect in large part because people couldn't agree
on what to include or not.

Mime
View raw message