Re: [boost] Re: [iostreams] Review Results -- Iostreams Accepted

From: Peter Dimov (pdimov_at_[hidden])
Date: 2004-10-04 12:48:00

Jeff Garland wrote:
> On Mon, 4 Oct 2004 18:01:44 +0300, Peter Dimov wrote
>> Jeff Garland wrote:
>>> Well now at least I think I can understand your obtuse comment from
>>> last week. Review managers in Boost organization are given much
>>> power and responsibility -- including the ability to decide against
>>> the majority. That's a fact, but it can be changed by petitioning
>>> the list for a change of process.
>>
>> What makes you think that review managers are allowed or encouraged
>> to decide against the majority?
>
> The review manager decides on the library. Nothing, in the process
> says the review manager must go by 'majority' rule -- it just needs
> to be based on the opinions of the reviewers.

Interesting. I have always thought that the review manager summarizes the 
review comments and decides on the library based on them. If that's not the 
case, why have a review?

The review manager can, of course, review the library himself. He is also 
given sufficient power to decide on issues where reviewers haven't been able 
to reach consensus, and he can tilt the tables slightly to overturn a 4-6 
vote (or similar).

That's how it happened in the past. Review managers that have posted review 
comments have usually been careful to mark them as such.

Never so far has a review manager decided against the majority.

> But anyway, my point was you are free to try and change the process
> thru discussion if you don't like it.

The process has worked well so far and needs no change, in my opinion.

>> It's different. It was fairly obvious that you decided that the
>> example was unacceptable for political reasons, and then tried to
>> justify you decision using additional non-political arguments (none
>> of which, in my opinion, hold water).
>
> Well, it's 'obvious to you', but that's not how it happened at all.
> I thought the example name was odd when I first saw it -- not
> compelling 'on-topic' for an iostreams library. By itself, not enough
> to worry about. Then a reviewer mentioned the political implications
> -- which got me thinking about it.  Could I think of any other
> example or documentation in boost that even remotely bordered on a
> political statement -- no.  If this did offend someone could it taint
> boost -- possibly.  Would I use this example if I were trying to get
> the library into the C++ standard -- nope.  Would the value of the
> example be less valuable if it were changed to be something
> apolitical -- no, it might even be improved.  Therefore it didn't
> seem like it was worth the risk.

I am not saying that this is not the correct decision. I am saying that the 
decision hasn't been supported by the majority of the reviewers as far as I 
can see.

> As for me injecting my political point of viewpoint (which you know
> absolutely nothing about), that just isn't the case.  I would have
> the same issue if the example was the 'Presidential Contender Filter'
> or 'President of France Filter'.

That's why I've been careful to not imply anything about what - if any - 
political viewpoint you might have. But maybe not careful enough.

>> Think about it this way. Someone _might potentially_ be offended by
>> the example, so you want it removed/changed. I am offended _right
>> now_ by your decision to remove/change the example, because I deem
>> it unjustified censorship.
>
> I'm sorry you are offended, but oh well. Anyway, I'm not going to
> apologize, because I don't believe I've done anything wrong.

This is not the point at all. Let's see if I can clarify:

No, you don't have to apologize to or accommodate people that are overly 
sensitive to censorship.

By the same logic, Boost doesn't have to apologize to or accommodate people 
that are overly sensitive to president speech disorder jokes.

No?

>> Then you'd be entirely within you
>> right to enforce it - but there would be no need to, because
>> developers will know not to include such content in their
>> submissions.
>
> Perhaps.  But even with updated guidelines, there is always the
> potential that a review manager will have to deal with a new and
> unique case not explicitly discussed by the guidelines. And no matter
> the language, the evaluation of
> what counts as a political tint might be hard.  Apparently there is
> some disagreement in this case...

Right. Which is why things like that are decided on a case by case basis by 
the reviewers. We only need procedures for the cases where this process 
doesn't work.

Apparently, you think that in this case the majority is wrong. Therefore, we 
need procedures.