Re: REMINDER: Release 3.6.0-alpha1 from master

Rainer Bielefeld píše v St 30. 05. 2012 v 11:57 +0200:
> Petr Mladek schrieb:
>
> > Note that alpha1 is there to find exactly this kind of problems. Of
> > course, it does not make sense to produce it if is would be completely
> > unusable. On the other hand, we do not want to push it on mirrors.
>
> Hi
>
> I see the problem that the alpha will not be tested enough, I (for 
> example) do not plan to use it for normal work (more or less), and so 
> the chance to find bugs decreases dramatically, because other (WIN?) 
> users also can't use the alpha.

I am not sure if there are good expectations set about the alpha/beta
builds. If they are perfectly usable and do not have bugs, we do not
need them and we could do final release immediately. There is always
risk to use them for normal work.

I am a bit confused. I am not sure why missing java would break you from
using the master build. IMHO, you do most of the work in Writer and Calc
These should not be affected by missing java that much. IMHO, java is
used there only for wizards.

I guess that you are more concerned about the random crashers. Well,
this is really bad problem. On the other hand, you are not able to
provide a good scenario to reproduce or backtrace. I am afraid that we
need to collect information from more people and alpha/beta builds are
there for this purpose.

Back to the expectation of the alpha build. We need to make sure that
beta1 will install/start and be basically usable on many systems. If it
crashes after half an hour of the work, it is pity but it is something
that might be expected from the first beta. Of course, we need to warn
people in the release notes and ask them to find a reproducible scenario
and report it into bugzilla...

What do you think? Is my view wrong in some way?

> > >    I still see the problem with a MinGW build
> > >    pull time 2012-05-28 04:23:12
> > > "Bug 50344 - Basic Macro will not run because of missing Java
> > >    <https://bugs.freedesktop.org/show_bug.cgi?id=50344>
> >
> > Ah, it is not that critical if it works
>
> Ah, some Misunderstanding. Java does not work anywhere. If that bug 
> would be "MinGW only" I would have added that key work in the summary 
> line. I only used MinGW for a quick test yesterday because that was the 
> most current Master available (we still do not have a MVSC Master after 
> 2012-05-26_02.43.54.)

Sure. MinGW is perfectly fine for testing. On the other hand, the normal
Windows build is what matters regarding release blockers.


Best Regards,
Petr