[Development] Build system for Qt 6

[Edward Welbourne]

On Tue, Oct 30, 2018 at 8:32 PM Pier Luigi Fiorini wrote:
>> From my point of view qbs is doomed as long as qmake's alive.

NIkolai Marchenko (30 October 2018 18:38) replied:
> I would much rather this abomination died instead.

You are not alone.  Unfortunately, the project has depended on it for
several years, with the result that we have a support commitment that we
can't simply abandon, much as we (especially those involved in
supporting it) would like to.  We have to make a graceful transition to
something else first.  The questions then are: what ? and: how soon can
we get to the point where we can put qmake's death on our road-maps ?

> I've switched to qbs when I got way too annoyed by how qmake does
> things and I've never been happier.

It's clear that QBS has great potential.  Unfortunately, the corporation
that initially supported its development does not see a way to monetise
it, so is no longer willing to fund its development.  It is available as
open source code that others can develop and maintain, but it's only
going to make it if others are willing to do most of the work.

In some sense, the history of mixed signals - TQtC continued development
but did little marketing and neglected important things like
documentation - is a symptom of it being an experiment.  Until
management had a full understanding of where this lead, they let
development explore the possibility that a better build system is worth
investing in.  Eventually they saw enough to make their decision and
decided to bail.  Some of us may not be persuaded that decision was for
the best; but I'm not used to writing business plans, much less
evaluating their viability, so I'm not going to try to tell them that
their decision harms the interests of their share-holders, which is the
only kind of argument I can expect the officers of a publicly-traded
corporation to care about.

Meanwhile, the Qt project gets to chose its build system.  If we have a
groundswell of support from outside TQtC that takes over development of
QBS and makes it capable of serving as that build system, I think quite
a lot of us shall be very happy to see it.  Otherwise, CMake is the only
candidate with a road-map to deliver tolerably soon, so there we go.
Painful as its syntax is (I've begun reviewing the work for it), it's
there, someone else is supporting it, and the expected time to the final
demise of qmake does look shorter than our other options.  Prove that
wrong by volunteering to work on QBS, or come to terms with the
non-ideal path we've landed on,

	Eddy.