Are we moving towards 1.35?
Hallo, I hope to be wrong, but I'm under the impression that, while everybody recognizes that issuing the libraries that have been detained by 1.34's delay is urgent, actual activity is stuck on a couple of strategic decisions. These are the structure of the Subversion repository and the new development process. Actually, I believe these are symptoms of a single meta-problem: how are strategic decisions taken within Boost? I can offer a few proposals: 1) S/he who does the job decides: unfair maybe but, hey, if you're interested, contribute! 2) The moderators decide: undemocratic, but then these are among the people that contributed the most and most continuously over the years; they gained the authority to make decisions on the field. 3) A poll is taken on the mailing list. 4) A non binding poll is taken on the mailing list, then the moderators decide. 5) The review process is used. In order to avoid the obvious bootstrap problem, I suggest that Doug Gregor and Beman Dawes get to choose how to reach a decision for the repository and the process, respectively. Cheers, Nicola Musatti
On Jun 25, 2007, at 6:51 AM, Nicola Musatti wrote:
These are the structure of the Subversion repository and the new development process. Actually, I believe these are symptoms of a single meta- problem: how are strategic decisions taken within Boost?
Yes, the moderators have discussed this issue recently.
I can offer a few proposals: 1) S/he who does the job decides: unfair maybe but, hey, if you're interested, contribute!
This is always going to be the default, because the only way to make a change is to show how the result is going to work.
2) The moderators decide: undemocratic, but then these are among the people that contributed the most and most continuously over the years; they gained the authority to make decisions on the field.
Boost has never been a "democracy" as such... for reviews, the review manager weighs input from the community and makes a decision.
3) A poll is taken on the mailing list.
Ick, no. This is precisely why we have review managers making decisions. We would not want our decision processes to "degrade to a popularity contest."
4) A non binding poll is taken on the mailing list, then the moderators decide.
5) The review process is used.
The moderators did discuss how to make decisions on larger pieces of infrastructure in Boost. Basically, some combination of 2, 4, and 5 seems like the best approach. We would have some public discussion of the proposed change (like we have for Subversion) or, if there is a strong technical component, an actual review (i.e., this is what one would do for a new build or documentation system). Then, the moderators would review the discussion and decide based on that. Think of it like a formal review, where the moderators (as a group) act as review manager. It seems like the most logical extension of the review process, applied to infrastructure and Boost-wide decisions. This isn't really a formal policy as such, but it seems to be the way we'll need to move forward. - Doug (Boost Moderator)
participants (2)
-
Douglas Gregor -
Nicola Musatti