Subject: Re: [boost] Review request for variant2, review manager wanted
From: Gottlob Frege (gottlobfrege_at_[hidden])
Date: 2019-03-03 17:44:30
On Mon, Feb 25, 2019 at 5:14 PM Niall Douglas via Boost
> >> (We actually use a completely renamed fork of Boost...
> > Whose idea was this, who set it up, who maintains it, what is
> > different from stock Boost, and what's wrong with using Boost as is
> > (as trillions of other developers already do)?
> It's actually not uncommon Vinnie in big megacorps. It's done to avoid
> the potential for implementation use of Boost symbol colliding with
> customer use of Boost. Each shop will have some poor sod who has to do
> the conversion from time to time, and maintain it for internal and
> sometimes external users. Tony van Eerd was that poor sod during my time
> at BlackBerry, for example.
Did I rename the namespace? I don't recall.
Sounds sad but reasonable.
> It'll usually have a bunch of nasty regex
> doing something like:
> namespace boost => namespace mycorp_boost_1_30
> boost:: => mycorp_boost_1_30::
> BOOST_ => MYCORP_BOOST_1_30_
> They usually rename the boost headers directory too, of course, to
> mycorp_boost_1_30/*, and all the library blob names.
> John Maddock maintains a tool which will do some of the renaming for
> you. But it doesn't rename macros, leaves some stuff in the global
> namespace due to ADL reasons, and some big megacorp likes to directly
> use Boost.Config macros to annotate their public functions and/or
> configure a custom Boost, so they need all the macros renamed too.
> Anyway, doing all the above is not uncommon. I've seen it a few times in
> my career now in various clients I've worked for. Usually with some
> ancient Boost version. And I've seen some big megacorp mix multiple
> different ancient versions of Boost into a single codebase, too, which
> is "safe" because nothing can clash. Yay.
> Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost