Re: [Banshee-List] gtk3 branch testing

["gnomeuser gmail com" <gnomeuser gmail com>]

2011/12/14 gnomeuser gmail com <gnomeuser gmail com>:
> 2011/12/13 Bertrand Lorentz <bertrand lorentz gmail com>:
>> On Tue, Dec 13, 2011 at 3:11 PM, Andres G. Aragoneses <knocte gmail com> wrote:
>>> On 12/13/2011 01:58 PM, gnomeuser gmail com wrote:
>>>>
>>>> 2011/12/13 Andres G. Aragoneses<knocte gmail com>:
>>>>>
>>>>> On 12/13/2011 01:32 PM, olivier dufour wrote:
>>>>>>
>>>>>> Hello all,
>>>>>>
>>>>>> We have raise a important step in gtk3 branch of banshee. It become
>>>>>> usable thanks to scroll-bar works now with the last git master of GTK#.
>>>>>> All graphical bugs have been removed thanks to a big work of Bertrand.
>>>>>>
>>>>>> But now we need test. Do not open a bug on bugzilla for it but if some
>>>>>> users can test it and do a complete feedback about test to track any
>>>>>> regression on mailing list it can be usefull to fix all remaining bugs
>>>>>> before merge it to main (which is not plan for now to be christal
>>>>>> clear).
>>>>>>
>>>>>> In order to test you need a mono parallel environment. To do that you
>>>>>> can use script from there :
>>>>>> https://github.com/nathanb/iws-snippets/tree/master/mono-install-scripts
>>>>>> or done it manually with this guid
>>>>>> http://www.mono-project.com/Parallel_Mono_Environments
>>>>>> After that, you need to build the last gtk# version from git master.
>>>>>> Maybe you will need few other library that you can found on
>>>>>> https://github.com/mono
>>>>>> So build all and build banshee gtk3 branch.
>>>>>> I know that some good guy as David will do so but more people we have to
>>>>>> test, less regression we will face later.
>>>>>> Bertrand will said when we have to open bug for gtk3 branch bug but , I
>>>>>> repeat, not for now. It is just to have more people who test to avoid
>>>>>> regression with the big merge.
>>>>>
>>>>>
>>>>>
>>>>> IMO the steps to do the testing are too complicated to do even for
>>>>> technical
>>>>> people. I think we should just merge the GTK3 branch as soon as possible
>>>>> to
>>>>> do a release. In the end we're still rolling out development releases,
>>>>> and
>>>>> this is exactly the purpose of this kind of releases: early feedback by
>>>>> users without the need of do the compilation themselves.
>>>>
>>>>
>>>> I tend to agree and as the, getting punished by OS X's strangeness on
>>>> a regular basis, guy I would like to have it merged quickly. If recent
>>>> history is any guide this is going to take an epic amount of time to
>>>> get working on OS X, let alone in a shippable state. However given
>>>> that the next release is set of the 21th (which my wife has graciously
>>>> agreed to delay the family vacation to allow me to do a build for), if
>>>> 2.3.3 is going to be GTK#3 based I would love for it to be merged
>>>> today. More time for packagers to get this all lined up and tested
>>>
>>>
>>> As olivier mentioned, we need to get other releases done before merging
>>> (gtk-sharp 3.0 beta the main one) so I guess today is impossible :)
>>>
>>> We could harass Mike Kestner to do the beta release asap, and if it's still
>>> not too late, merge gtk3 before the 21th...
>>>
>>> But my feeling is that it's maybe rushing a bit. What we may want to do is
>>> get the gtksharp and gudevsharp releases out before the 21th, and then just
>>> after releasing 2.3.3, merge gtk3 branch into master.
>>>
>>> Let's see what Bertrand thinks!
>>
>> Gee, and I still wonder how I ended up apparently running this show... :)
>
> It's your sexy trust inspiring beard and enhancing accent :)
>
> Regardless I did some preliminary work on packaging this.
>
> Instant problems:
> librsvg doesn't compile out of the box against a gtk3 stack on OS X
> (will look into this but it looks to be some ASM stuff which is going
> to be.. fun).
>
> Theming, Murrine requires gtk2 and tango-icon-theme requires librsvg
> so we are going to look like crap for now I suspect.
>
> gst-plugins-good attempts to compile cairooverlay which doesn't work
> against gtk3 (so no ogg, flac and such).
>
> mono-addins requires gtk-sharp2. This sems like porting will be
> required so volunteers for that would be fantastic.

Bertrand helpfully suggested that --disable-gui takes care of this.
However now ige-mac-integration (or rather the new replacement
gtk-mac-integration) has posed a blocker by failing to compile. This
is becoming fun.

>
> So there is still plenty to do on the packaging side to make this
> work. I already had to package what seems like half of X to make the
> required cairo to compile to enable introspection in gtk3.
>
> It is not as big a job as I expected but getting things to even a
> place where I can distribute a test image is going to require some
> help.
>
> Of course gudev-sharp isn't relevant to OS X so I get off easy on that one.
>
> -David
>
>>
>> Well, I have two pull requests pending for gtk-sharp :
>> https://github.com/mono/gtk-sharp/pull/31
>> https://github.com/mono/gtk-sharp/pull/32
>>
>> As you can see, they are both related to memory management, and I
>> believe they fix some leaks, but could cause some if I'm doing things
>> wrong. I guess Mike is being careful, as, since I opened those, he has
>> merged some other, simpler, pull requests.
>>
>> There are probably some other issues lurking around in gtk-sharp and
>> our gtk3 branch, but I think those would be crashes or "being ugly"
>> issues, not "eating your data" issues.
>> By the way, speaking about being ugly, I'd like to get rid of the
>> border around the toolbar, and experiment with the "primary-toolbar"
>> style.
>>
>> While it might be OK for development releases to depend on git
>> snapshots, I don't think it's reasonable to have this for a major
>> stable release. So the question is: will our dependencies, gtk-sharp
>> is particular, be ready for 2.4, which is scheduled for March 21st
>> (and preferably for our feature freeze, on February 15th).
>>
>> Please keep in mind that, once the gtk3 branch is merged in git
>> master, backing out later on would be... messy. The diff between the
>> two branches for both Banshee and Hyena is about 2000 lines added,
>> 3500 lines deleted.
>>
>> So I don't think there's a good reason to rush the merge before the
>> 2.3.3 release. My suggestion would be to re-evaluate the situation in
>> early January, and see how we proceed then.
>> Who knows what Santa Claus and the New Year might bring ! :)
>>
>> In the meantime, build, test, report bugs, write code !
>>
>> --
>> Bertrand Lorentz
>> _______________________________________________
>> banshee-list mailing list
>> banshee-list gnome org
>> http://mail.gnome.org/mailman/listinfo/banshee-list  (unsubscribe here)