Boost logo

Boost :

Subject: Re: [boost] [process] [sorting] [singularity] [compute] [others] Who needs review manager?
From: Robert Ramey (ramey_at_[hidden])
Date: 2014-09-14 14:05:30

Rob Stewart-6 wrote
>>> judge - review manager
>>> jury - reviewers.
> There's a problem with the analogy because we do expect the Review Manager
> to ensure the case is ready for court and the defense is fully prepared,
> so to speak.

where's the problem? In the legal world the judge makes sure that the
parties have time to prepare, will grant them more time if he believe its
justified, rule on what evidence will be admitted etc, etc. I decides,
but generally doesn't participate in the argument. It's an explicit attempt
to separate the roles of advocate from those of judge.

This analogy has worked well for us. I don't remember any case where
authors of rejected libraries have blamed the reviewer for unfair treatment.
To me its an indication that someone "got it right". But of course there
are still opportunities for improvement. I would like to see the following
rule added.

A review can't proceed until there are N reviews already prepared and
sitting in the incubator. I'm thinking a reasonable N would be 4?

This would mean
a) more people have incentive to use the incubator.
b) the preliminary effort the reviewer has to go through would be a little
c) the incubator has a low bar of non-controversial requirements. A library
can't pass this bar is not really reviewable.
I think this tiny addition would make the reviewers job a lot easier and
save wasted effort with no discernible downside.

One more thing. reviewers recommend that the library be rejected, accepted,
or accepted subject to conditions. I think lots of people see this as a
"vote". It suggests that the decision is made by vote. This is not at all
true and stating this is very misleading. One of the key points of the
review process is that one person weighs all the considerations and is
personally responsible for making the decision. It's a heavier burden than
just counting votes. Its clear from the review results emails that the
reviewers take this responsibility seriously and that Boost has gained
mightily from this. I would like to discourage reviewers from referring to
their recommendations as "my vote" rather than "my recommendation". I
realize that I'm blowing this out of proportion, and I can't really
understand myself why I'm picking on such a nit. So please forgive me for
indulging in a very personal quirk here.

Robert Ramey

View this message in context:
Sent from the Boost - Dev mailing list archive at

Boost list run by bdawes at, gregod at, cpdaniel at, john at