[oe] [meta-qt5] Update to 5.1.1?

[Martin Jansa]

On Tue, Oct 08, 2013 at 02:50:59PM +0100, Laszlo Papp wrote:
> On Tue, Oct 8, 2013 at 2:39 PM, Martin Jansa <martin.jansa at gmail.com> wrote:
> 
> > On Tue, Oct 08, 2013 at 02:30:05PM +0100, Laszlo Papp wrote:
> > > On Tue, Oct 8, 2013 at 2:26 PM, Denys Dmytriyenko <denis at denix.org>
> > wrote:
> > > > Since Dora is a branch of meta-qt5, I don't understand your statements
> > > > about
> > > > being orthogonal and not integrated...
> > > >
> > >
> > > You will be able to use *any* meta-qt5 on top of Dora, just as you can
> > use
> > > meta-qt5 on top of denzil nowadays. The "dora branching" sounds like red
> > > herring to me. The dependencies in Dora will be good enough for 1-2
> > years,
> > > if not much more.
> >
> > That's not true, we were just lucky that there weren't any incompatible
> > changes in oe-core, but as soon as 1.6. brings e.g. new cmake, master
> > branch will work only with master and dora only with dora (because of
> > .bbappend).
> >
> 
> I am not sure who it is not true. Qt has been working for me, and even if
> there are small quirks, they community can and will provide changes. Even
> if they could not, they can use .bbappend, and custom changes for any
> recipes. That is the strength of open embedded after all. It is very
> unlikely that everyone will always like every package.

That's basically what I said, if they use unsupported combination, they
can expect that they will maybe need to resolve some quirks on their
own.

>  So the same branch names are the only supported combination, if you want
> 
> > to use meta-qt5/master with oe-core/dylan then you're on your own. It
> > works now, but only because we're lucky.
> >
> 
> IMO, this is unreasonable. It will cause additional overhead and looking at
> the manpower behind this layer, I am not really sure it is worth it.
> Quality should be over quantity. Maintaining several branches rather than
> just having one and well-qualified branch for starter would be more
> fruitful in my opinion.

Everything goes to master first, bugfixes can go also to release
branches which track older oe-core releases, the same rules which apply
to meta-oe and other layers apply here.

> > > 5.2 comes with several fundamental improvements. I would not personally
> > > recommend Qt 5.1 to anyone in about two months time when the release can
> > > happen. :)
> >
> > There are companies still using 5.0.0 just because they fear to upgrade
> > or because newer qt versions were released to late for their release
> > cycle of real hw products. I agree that it would be nice for everybody
> > to just use latest, but there are still meta-qt5 users which cannot do
> > that.
> >
> 
> So what? They can always use bbappend, own packages, old branches, etc.
> That should be in no way obstacle for new innovation.

Yes, that's why meta-qt5 will exist in "old branch" for each oe-core
release and such branches will get only bug fixes, because people who
cannot upgrade oe-core to latest probably don't want to upgrade qt5 to
latest as well.

-- 
Martin 'JaMa' Jansa     jabber: Martin.Jansa at gmail.com
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 198 bytes
Desc: Digital signature
URL: <http://lists.openembedded.org/pipermail/openembedded-devel/attachments/20131008/e3354adc/attachment-0002.sig>