Boost logo

Boost :

Subject: Re: [boost] AFIO review response
From: Niall Douglas (s_sourceforge_at_[hidden])
Date: 2015-08-30 12:24:46

On 29 Aug 2015 at 23:24, Pete Bartlett wrote:

> Robert wrote:
> >I hope [Ahmed] can find enough time to [manage the review]. I'm sure that
> it's more than he thought he was signing up for.
> In my view Ahmed's job is _potentially_ quite easy:

There are about five main topics of debate by my count. I think a
compromise or consensus is approaching in each, and that should make
it easier to summarise.

> I think Niall is a smart
> and big enough guy to realise that even if he applies a discount factor to
> those reviews that he thinks are overly-hostile due to a personality clash,
> the general consensus is still not strong enough to an "accept" at this
> time, and therefore a withdrawal is appropriate. He could come back with a
> plan - "here's what I am going to change, here's what I am going to keep the
> same" and hopefully get some immediate feedback. Maybe some will say "sorry,
> you not changing X is a showstopper" and Niall can choose whether he wants
> to pursue with the changes. If he wants to pursue, we re-review.

Exactly what I envisioned. The review to date has been very useful,
and is exactly what I needed at this stage.

> Personally I like the library and would be minded to accept regardless of
> whether Niall chose to keep his "shared" style or switch to the "standard"
> style outlined by Thomas Heller (and apologies if I am pushing anyone's
> buttons by calling it "standard"). That seems to be the only substantive
> design issue - the rest - the namespaces, the #if 0, parts of the
> documentation, the irrelevant v1, is basically bike-shedding stuff that
> could be changed after an accept.

I absolutely agree. And I think we may be approaching a compromise in
the Heller thread, and even that alone is but a few hours of work to

To date assuming we reach resolutions as they currently appear to be
heading, I count less than 40 hours of work needed to repair anything
code related that has been raised in this review so far. The only
real showstopper is the documentation, and I suspect another 40 hours
would fix 80% of the most reported problems. That's just a few months
of low hanging fruit to pick. It's less than I expected.

> Finally whilst I'm up on the soapbox, I'd urge you, Niall, to just be a bit
> more careful with how you come across sometimes. What I view as enthusiasm
> for code and love for talking about code and coding, does seems to comes
> across less well with others. E.g. are you offending others by appearing to
> speak for them or implicitly denigrating their efforts by highlighting your
> own? [I am sure this sounds like you are having your hard work such as the
> "best practices handbook" thrown back in your face, but honestly I am only
> trying to improve the environment in which any further review might take
> place]

I do take exception with negativity for negativity's sake true. I
also believe that collectively acting as if developing Boost
libraries is effortless and does not involve major sacrifices is the
Ayn Rand-ian hero taken too far. If you start from gratitude when
criticising, you'll get nothing but positivity in return from me.

All this said, I recognise I have been unusually tetchy since
February this year when this uber work pace began, and I acknowledge
I could have behaved a lot better on many occasions, particularly
with regard to unenforced poor judgement. The good news is in just
three weeks from now this burst of work is over, and I will be
returning to my standard 50 hour week for the foreseeable future
which includes most of 2016. I already have a fitness and
physiotherapy schedule planned for as soon as I return from CppCon as
I have lost a lot of fitness and gained a fair bit of weight since
Feburary. I'm very much looking forward to that.

> Nice job with AFIO!

Thanks Pete. Much appreciated.


ned Productions Limited Consulting

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