Subject: Re: [boost] [review queue] What to do about the library review queue?
From: Niall Douglas (s_sourceforge_at_[hidden])
Date: 2017-03-14 13:35:29
> I personally hope none of the suggestions 1, 2, 3, or 4, are
> implemented. I really worry about this desire to bring monetary value
> or payment into the Boost review process.
> I also have some concerns when I see the same person put forward
> similar ideas that all revolve around paying or hiring individuals.
Every single major open source org has eventually ended up having to pay
people to do the admin which keeps that org working if they wish to keep
growing. And that's an empirical hard fact.
Boost can decide to not do that, and to date it has decided to not do
that. All empirical evidence suggests that that decision will gum up
growth and put a natural cap on Boost's size and relevance to latest C++.
A few years ago Boost had a problem of no new libraries at all in three
years. That problem has been fixed, and I'll immediately admit surprise
tinged with gratitude that it didn't require paying people to fix it.
An ongoing current problem is lack of maintenance. A system of corporate
sponsorship of maintenance has helped address that, and in some places
it's worked well, but not in others. But I'll also admit surprise tinged
with gratitude that anything non-monetary worked here at all.
A big surprise to me is how much has been done on automated testing
without paying people for it (though we did pay for some of their
hardware and in some cases for renting infrastructure). My hat is off in
thanks to those in question. Thank you.
So, one could entirely argue and with good supporting evidence that you
don't need to pay people to do this stuff, and you would be correct.
However that does not mean that if you did pay people, that it wouldn't
have gone a lot better again. And there is a much stronger argument that
the fact I raised payment for stuff and people reacted violently against
it helped get people to volunteer to solve the problem. So in a way,
it's all good and it all helps.
> Niall, I understand you have good intentions, and I (like many others)
> appreciate your administration of GSOC, but the repetition of these
> ideas looks frighteningly like you want to address any employment or
> financial needs that you have by getting Boost to hire you for more
> things. Maybe I'm drawing too many conclusions from your blog
Heh. Funnily enough I'm coming towards the end of my active
participation in C++, you've got about a year left of me being annoying
before I step back majorly. Even this year you won't be seeing me at
CppCon this year as I begin to step back.
(To explain, we've been having children since 2013 for which my wife had
to give up her career and I was the sole earner. She wishes to return to
her career soon, and for which all my non-work time will be needed to
support her, so no more outside-of-work coding for me for a few years
until she's reestablished and back to earning. I'll be resigning from
all things I volunteer for including all mailing lists, and doing
nothing but basic maintenance on my open source libraries, no new code
nor new features)
You are right though that in *ideological* terms I think Boost and the
C++ Standard Foundation ought to hire people to work on the wider C++
ecosystem, just as other major programming languages do. Whether that
should be me or someone else should be the result of a competitive
public tendering process.
> I'm curious: The libraries that too niche or esoteric for any review
> manager to be interested (or reviewers to come forward) - to which
> libraries are you referring?
Safe Numerics is an excellent example. I've never used anything like
such a thing, and despite it looking to me like a great library (and I
did look over it in depth including its source code), I was aware enough
of how little I knew about that domain. I definitely could not have
review managed that library, I don't know enough about that domain at
all to have a useful opinion.
> In any case, I look forward to the mechanics of Boost being improved.
> I hope they're improved by better means that do not involve solving
> someone's employment needs.
Nobody is going to be employed permanently on bits of piece work, it's
an awful life not knowing where rent will come from. Any of the remote
working consultants I know prefer nice, long contracts where available,
ideally six months or a year or longer.
That said, when you're between those long contracts and during when
you're negotiating new contracts (which can take two months like the one
I'm negotiating right now has), if there were bits of small piece work
available, I'm sure I and many others in this profession would put a bid
in to fill the short gap at a very discounted hourly rate. That would be
excellent value for money for the C++ ecosystem, and therefore a very
rational system to establish.
-- ned Productions Limited Consulting http://www.nedproductions.biz/ http://ie.linkedin.com/in/nialldouglas/