On 11/10/24 10:20 AM, Peter Dimov via Boost wrote:
A review manager should act like an impartial judge through the process, not like the library’s lawyer.
When I first joined the Boost mailing list, years ago, I also thought that the review manager is supposed to impartially reflect the community opinion, as expressed in the form of formal reviews.
But it turned out that I was wrong. The review manager's role was, as I later figured out by observing the process, to decide whether the library should be accepted, and the reviews were helping him with this, rather than deciding for him.
It was common, for instance, for the review manager to actually write a review, usually prefaced with "this is my review of the library which I submit independently of my role as a review manager."
This doesn't make much sense if the review manager is only supposed to impartially tally votes.
I didn't particularly agree then with that process - for me the review manager was, in fact, supposed to tally votes - but the process was what it was, and it worked.
I think this touches on a severly under appreciated aspect of the Boost Review process. Boost is not a democracy. At least not a pure democracy. The review process is not in any way a democratic process. Reviews are not votes. Libraries are acceptance/rejection are the sole within the authority of the review manager. Just to over state the point - the Boost Review process is not like congress - just tallying the votes. It's more like the Supreme Court - does the library meet the accepted standards for a acceptance as a Boost library. This is resolved by various advocates presenting their cases, criticizing other advocates cases, presenting facts, logical arguments etc. The Review manager's job is we weigh all this and conclude whether or not the submitted library meets Boost standards. If concensus can't be reached - the process ends without a resolution. Any process is going to produce irreconcilable differences. In the "congress" method, concensus/compromise is achieved through negotiation and concensions among competeing parties. In the "supreme court" method, a decisions reached unambiguously. There may be conditions or not. With either method there will be "winners" and "losers". So it's not about everyone being happy at the end of the day. It's about getting the best outcome. I firmly believe a) The best engineering designs result in simpler, more useful, easier to maintain, longer lasting, and more beautiful designs. E.G. the Golden Gate Bridge; great design. The Oakland Bay Bridge: WTF? b) Having a single simple definable goal - getting people across the bay leads to a better design than having multiple goals which could conflict - build a bridge which gets people across the bay and promotes the local economy, which respecting Land rights from people long dead and promotes special opportunities for marginalized groups, etc.... c) that the compromise, consensus, in engineering endeavors leads to poorer design. d) that this goal is best achieved by delegating the decision to the single most capable person in the area that we can find. e) This person will for better or worse be personally responsable for making the decision, defending the results, dealing with future criticism. His name will be public. This is not a job for everyone. Contrast the Boost Review process an it's results (for better or worse) with that of the C++ standards committee. Consider for example the example of asyncronous I/O. Have fun with this. Robert Ramey
_______________________________________________ Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost