[Scummvm-devel] 1.2.0 release discussion (was Re: GSoC'10 is over. Great success!)

[John Willis]

> On 24 August 2010 14:21, Filippos Karapetis <bluegr at gmail.com> wrote:
> > There are many things to be added for version 1.2.0, including the
> > GSoC code, and the new engines (i.e. SCI, Hugo, AGOS/Personal
> > Nightmare, Fascination, Big Red Adventure....) Is any of these going
> > to be added to 1.2.0? I'm guessing that Personal Nightmare will be
> > added, not sure about the others. Are we adding any of these, or will
> > they wait for 1.3.0?
> I would say this.
>
> It is pretty safe to integrate 2 GSoC tasks, those are decompiler and
testbed
> engine. Both are completely independent.

I would tend to agree, get a clean release out the door soon'ish and then
deal with the GSoC merge of the 2 remaining projects and the instability
window after the branch.
 
> The things which are safe to announce:
>   * Translation support
>   * Improved GUI
>   * Androind and Dingux ports

Cool, 

I would also like to consider adding 2 additional ports in time for the
release if it can be accommodated (they have not been committed yet but both
are in a pretty solid shape).

One for the OpenPandora console and one for the GPH Caanoo (the successor to
the GP2X Wiz, same code with some very minor changes and a different build
toolchain). 

The Caanoo port is held up by the fact I am waiting for shipping hardware to
re-test it on (should get it in a week or so), the OpenPandora port has been
held up by lack of free time but I plan to get it committed to trunk in a
few days.
 
> As of the version numbering. I see no difference is 1.2.0b followed by
1.2.0,
> or 1.2.0 followed by 1.2.1. It is exactly the same thing release-wise. If
you
> guys feel better with betas, I am completely OK with it.

Psychologically something about the '-rc' and then full release sits better
with me but your right, it's all down to semantics. The release process is
the same regardless.

> The main concern for me now is SCI engine. The testing although being
> fruitful and effective became a bit of load to the developers. I do not
want
> to worsen the situation or push low quality engine to release.  It is up
to you
> guys to tell me whether you are ready or not. We may decide to save SCI
for
> 1.3.0.
> 
> Currently our release got slipped by a month, i.e. usually we had it on
last
> week of August. If we start now, we will have it at mid of September as
the
> fastest.

I would think late Sept. is more realistic.

> Another subject to discuss is possible skipping of testing for particular
> games. What I am thinking about is to analyze the commit history for each
> engine and decide between those engines which stability was not altered
> since the last release (i.e. no significant commits), and those which were
> changed or are new. Then skip testing altogether for stable engines. The
> number of games is really increasing, and without the event recorder it
> becomes more and more hard to perform a full-blown pre-release testing.
> What do you guys think?

Seems a solid idea to me.

Regards,

John