[Qt-creator] Planning Qt Creator 2.8 + 0.1

[Poenitz Andre]

Hi all.

With 2.8 beta out of the door it's not yet time to call the release done, but
we can already spend a few brain cycles on what should happen after 2.8 final.

A few ideas are already sort of fixed, a few completely open for discussion,
some need feedback from users and the developer crowd.

1. Time line: We'd like to stick to the usual 4-month cycle, putting the
target date for the Final somewhere to November. Reason: The current cycle
setup works well, no need to change. Also November is generally a rather nice
month for releases, due to its distance to holiday seasons.

2. 2.8 + 0.1 makes 3.0, as we use base-9 for version numbers ;-). More 
seriously, we'd like to make the point that with Android and iOS support we 
have a new "phase". At the same time we'd like to stiffen the rules on core
compatibility to make it easier for 3rdparty plugin developers to keep their
plugins working. Current thinking is to aim at source and binary compatibility
within a minor series (i.e. 3.0 and 3.0.1 could be interchanged, but not, say,
3.0 and 3.1).

3. We will focus on building of Creator itself with Qt 5. The plan is to keep it
compilable as long as painlessly possible with Qt 4. I think by then (almost a
year after Qt 5 came out) the point has been made that it is possible to keep
a code base of the size of Creator without much pain compilable with Qt 4 and
Qt 5 at the same time. It's time to get access to the Qt 5-only goodies now.
This would also make it possible to upstream generally useful parts from
Creator into Qt again which is currently hindered by the "don't play with Qt 4" 
policies.

4. "committed" contents: Essentially tying up loose ends. There are obvious
gaps in Android and iOS support that needs to be closed, and we have a couple
of other "in progress" or currently "experimental" items (multi-monitor
support, lldb debugger support) that should be finished.

5. "committed" maintenance: In preparation of the potential compatibility
promises the core interfaces need some auditing, and possibly re-shuffling
and "real" documentation. In addition there should be general performance
audit/profiling including fixing the most glaring issues that will come up.

Not-so-fixed:

6. Use of C++11. There are quite a few goodies in the language nowadays
that look very interesting. We should identify a few ones that would really
help us, and double check with the compilers we need to support which
ones we could actually use, and then decide on whether we should, or not.
For that, input on which compilers people actually use, and which they
really need to use to compile Creator would be beneficial.

7. wip/clang: Progress would be highly desirable. It's unlikely though, to get
this anywhere close "production ready" in this period, so the current thinking
is to bring it in a state where it's user-configurable, and perhaps have that
configurable per-language, so that it can get some exposure in the Real
World for, say, C projects, where the performance impact is less visible.

Andre'