[OE-core] [PATCH 3/3] webkitgtk: update to 2.24.1

richard.purdie at linuxfoundation.org richard.purdie at linuxfoundation.org
Mon May 13 13:46:18 UTC 2019


On Mon, 2019-05-13 at 15:39 +0200, Alexander Kanavin wrote:
> On Mon, 13 May 2019 at 13:36, Richard Purdie
> <richard.purdie at linuxfoundation.org> wrote:
> > On Wed, 2019-05-08 at 15:54 +0200, Alexander Kanavin wrote:
> > > Add a patch to address a missing format string error.
> > > 
> > > Disable the recipe on 32 bit x86, as SSE2 is now required
> > > unconditionally.
> > 
> > Unfortunately this isn't quite that simple since just disabling it
> > triggers:
> > 
> > https://autobuilder.yoctoproject.org/typhoon/#/builders/85/builds/445
> > 
> > (errors in world builds)
> > 
> > It might be time to optimise qemux86 slightly more which would help
> > avoid this?
> 
> Yes, we should certainly move qemux86 to something less vintage than
> the original Pentium CPU. I'll make a patch.

Sounds good thanks.

> Another issue is 'packagegroup-self-hosted' including epiphany, and
> build-appliance-image pulling that packagegroup in. Tracing back the
> history, it seems that epiphany is there because long time ago Hob UI
> made use of a 'web browser' for some tasks. With Hob long gone, and
> its replacement Toaster accessible across the network (or from outside
> the VM), I think there is not any more a use case for including a web
> browser into the image?

I agree it shouldn't be in self hosted which should be more minimal. I
think we do need a webbrowser image which we test *somewhere* and this
ties into the discussion we had in the bug triage meeting regarding bug
12991.

Basically "build-appliance" was designed to test whether toaster/hob
could rebuild the build-appliance image (and hence be self hosting).
I'd like to see this split up so that the self hosting is commandline
and we have a web browser test somewhere else.

An added complication is that the test also tests VMWare :/. Tesing
multiple things with one operation has its drawbacks...

Cheers,

Richard




More information about the Openembedded-core mailing list