[Sugar-devel] depending on introspection

[Tomeu Vizoso]

On Wed, Jun 16, 2010 at 16:28, Peter Robinson <pbrobinson at gmail.com> wrote:
> On Wed, Jun 16, 2010 at 4:00 PM, Tomeu Vizoso
> <tomeu.vizoso at collabora.co.uk> wrote:
>> On Wed, Jun 16, 2010 at 15:17, Peter Robinson <pbrobinson at gmail.com> wrote:
>>> On Wed, Jun 16, 2010 at 11:27 AM, Tomeu Vizoso
>>> <tomeu.vizoso at collabora.co.uk> wrote:
>>>> Hi,
>>>>
>>>> anybody has thoughts about the convenience (or not) of making Sugar
>>>> depend on the introspection stack in GNOME 3.0?
>>>>
>>>> The biggest practical downside will be that Sugar 0.90 will only run
>>>> on next-cycle distros (Fedora 14, Ubuntu Maverick, etc) unless people
>>>> backport a lot of other packages (not recommended nor likely).
>>>>
>>>> The upsides include: gradually dropping static bindings which are
>>>> generally unmaintained, less memory use, less cpu usage during
>>>> startup, access to new APIs such as GSettings and Telepathy-GLib.
>>>
>>> I think its inevitable that we go that route. The only suggestion I
>>> would ask about is the requirements to be able to support it on
>>> RHEL/CentOS 6. Fedora 12/13 already had some introspection support so
>>> is it much different in requirements to those in the initial
>>> implementaiton as I think for long running suppor I believe there is a
>>> desire to be able to use that platform. If they are new packages we
>>> can easily add them into EPEL so that´s not too much of an issue, the
>>> issue comes is if the upstream package is in RHEL mainline that we
>>> can´t duplicate it.
>>
>> Hmm, I think we would need newer glib and pygobject. Would that be possible?
>>
>> If so, then by staying with gtk+ 2.0 versions of all the libraries we
>> could run on RHEL, but from what I have read in desktop-devel, not all
>> maintainers are keen on doing that.
>>
>> How bad would be the consequences of Sugar 0.90 requiring components
>> only in GNOME 3.x?
>
> I´m not sure to be honest. I think we´ll only be able to tell that
> properly once RHEL-6 is out, it is anyone´s guess as to what their
> plan is with the desktop side of it. Being desktop its possible that
> if its not supported initially that support will be added later as it
> doesn´t impact the server side so much. Its something worth
> considering but I don´t believe it should be the only consideration.

Just asked in #fedora-desktop:

<tomeu> I guess we cannot expect a complete introspection stack in RHEL?
<walters> tomeu: nope, unfortunately
<walters> but for fedora 14 ideally it's essentially finished

Colin cares mostly about gjs, but my impression of PyGI is that it
will be feature complete in F14 as well.

Regards,

Tomeu

> Peter
> _______________________________________________
> Sugar-devel mailing list
> Sugar-devel at lists.sugarlabs.org
> http://lists.sugarlabs.org/listinfo/sugar-devel
>