[vlc-devel] One week to 0.9.0-test1

[Remi Denis-Courmont]

On Tue, 24 Jun 2008 21:19:06 -0700, Jean-Baptiste Kempf <jb at videolan.org>
wrote:
>> The biggest issue at the moment appears to be the OSX interface. Not a
>> big surprise with the OSX developers away.
> What issue are you talking about? I see the preferences on my list, but
> nothing else. (I know it is a big stuff, I will look at it next week).
> (#1 and following, IIRC)

According to Trac, the input handling is broken, which is a total
no-starter.

> - Qt4 embedded video is better [68cb77], [6ec764] and [8b43c0] but not
>   completly fixed (crashes on exit). Maybe related to other Vout
issues...

In other words, it's completely unusable.

> - Qt4 fullscreen controller works in EV, but doesn't else and there are
>   still stability issues (Lukas, can you look at it please?)

Basically, it crashes all the time, which means it's unusable as far as I
am concerned. If it's not fixed, there is no option but to leave it out.
Same thing for the embedded video.

> ** Vout don't close/quit on stop like they were used to. Is it a bug or a
>    feature? (I don't remember)

I have no problem with the vout on Quit. I have not tried Stop lately, but
it would be a regression in the input core.

> ** Skins seem to be broken on windows. Can someone confirm?

I don't know.



Generally, as the project goes forward, gets more features, and older
contributors leave, there is no way we can forever preserve every single
feature ever added forever. I happen to believe that open-source release
management is all about prioritizing severe bugs and patches. It is not
about managing features, and definitely not about managing feature
requests. If an issue is not a complete showstopper, and nobody is willing
or able to work on, it's got to be ignored. Otherwise, nothing will ever
get released. Ever.

> I think Qt4 Iconset and Contribs can wait after test1, no?
> Shrinking the size of the windows installer can wait too.

Contribs are not part of the release per se. You can change them anytime.
It's all about ensuring that the contribs in any specific _binary_ build
are usable. It's pretty much the same with the installers.

> I would just like to ask a question: what is our road?
> (test1, test2, rc* )? (test1, test2, test*)?
> It would be nice to have a way for the update system to work. (I don't
> think rc > test, for example).

That's why you should do either "-preX" -> "-rcX" -> "", *or* "-testX" ->
"". Not a mix of these..

I think the basic idea is that by -test1, all known blockers are gone (or
corresponding features postponed to the next release), and then there is
only bugfixing new issues, and non-source code problems (installers,
contribs, translations, images...).

In any case, we must have one last prerelease with a certain amount of time
to the release for translators.

-- 
Rémi Denis-Courmont
http://www.remlab.net