[mythtv] Announcement: Master/0.28 is moving to QT 5.2

[Gary Buhrmaster]

On Wed, Mar 4, 2015 at 8:19 PM, George Nassas <gnassas at mac.com> wrote:

I am not a dev, but I have stayed at a Holiday Inn Express...

> Could you set expectations for people with things like what you mean by
> “ready” and should we expect current master to be usable for production or
> is it a total crap shoot at this point.

On a scale of 0x0 to 0x7, it is likely a 0x5.  But master has always
been wibbly wobbly regarding "ready".  It is called "unstable/development"
for a reason.  Mostly it is good.  Sometimes there are rough spots.
The rough spots get fixed, but you might get unlucky with the
timing and pull at just the wrong time, and encounter some serious
issues.  Caveat Emptor.

In particular, right now, there were/are some reported segfaults that
are actually due to real code bugs that (apparently) in Qt4 happened
to work (most of the time?) but in Qt5, not so much.  I would not
be surprised to know that some of the random segfaults that have
been reported over the years are related to this (timing, and race
conditions, are everything).  There are likely more such in the
code.  Testing by users of the various code paths are a useful
tool.

> Also, would you feel master to be solid if you’re on qt4 libraries?

It is intended that RSN new code will be added to utilize
the (new) Qt5 features/constructs such that master will
not build on Qt4.  There will be no (intentional) effort to
insure that master builds on a Qt4 base (previously you
may find code in master/fixes that was intended to work
on either version of Qt).  I would plan accordingly.  And
remember, the bugs being identified with Qt5 are real
bugs.  You might have gotten lucky with Qt4, but the
bugs are real.