[OE-core] [PATCH 0/1] Upgrade gcc 4.6.0 to latest on FSF 4.6 branch

Richard Purdie richard.purdie at linuxfoundation.org
Tue Jun 14 12:37:03 UTC 2011


On Tue, 2011-06-14 at 10:28 +0100, Phil Blundell wrote:
> On Mon, 2011-06-13 at 10:12 -0700, Khem Raj wrote:
> > On Mon, Jun 13, 2011 at 1:51 AM, Phil Blundell <pb at pbcl.net> wrote:
> > > On Sun, 2011-06-12 at 21:03 -0700, Khem Raj wrote:
> > >> This patch brings in new patches from gcc 4.6 FSF branch
> > >> And refreshes the headers of existing backported patches
> > >> to not have git patch numbers in comments
> > >>
> > >> I am not sending the patch itself to mailing list due to its
> > >> large size so please review it on the contrib tree itself
> > >>
> > >
> > > Would we not be better off just pulling the tip of the 4.6 branch from
> > > FSF SVN, rather than having to keep all these patches in git?
> > >
> > 
> > there is dislike for this approach in oe-core. As the release point is preferred
> > I suggested to drop the minor release number and call the recipes 4.6
> > and use SVN
> > REVs to track the recipe updates but it did not fly :)
> 
> Where does that dislike come from?  Koen did make a comment about having
> liked svn checkouts for 4.5 "very, very much" but I couldn't quite
> figure out whether he was being sarcastic or not and, if so, what
> exactly his objection was.
>
> I could understand there being a preference for individual patchsets if
> we were just going to cherry-pick carefully selected bugfixes from the
> branch and patch them in.  But, if we're going to take the approach of
> just importing everything from the branch en masse, it seems like
> keeping them as patches is just making more work for ourselves.
> 
> We're using svn checkouts for eglibc, which seems to be working well
> enough and hasn't provoked any particular outrage that I noticed.

I think it was my dislike that Khem is referring to. I was under the
impression that we were going to be more selective that just taking
everything (e.g. the translation updates probably aren't essential to
us).

I realise its easier to just take everything though and if we are going
to do that it probably does make sense to use svn directly. I'll take a
patch to do that.

Cheers,

Richard






More information about the Openembedded-core mailing list