[OE-core] [PATCH 00/18] Provide list of deployment artifacts

Ed Bartosh ed.bartosh at linux.intel.com
Tue Aug 30 11:21:59 UTC 2016


On Tue, Aug 30, 2016 at 07:58:31AM -0300, Otavio Salvador wrote:
> On Tue, Aug 30, 2016 at 7:35 AM, Ed Bartosh <ed.bartosh at linux.intel.com> wrote:
> > On Tue, Aug 30, 2016 at 07:21:20AM -0300, Otavio Salvador wrote:
> >> On Tue, Aug 30, 2016 at 6:29 AM, Ed Bartosh <ed.bartosh at linux.intel.com> wrote:
> >> > Hi,
> >> >
> >> > This is a fix for Bug #9869 - Provide a per-target manifest of files which were, or would have been, produced
> >> >
> >> > The list of artifacts produced by deployment tasks (do_deploy, do_image_complete and do_populate_sdk[_ext] is
> >> > obtained from sstate manifests and fired as a TaskArtifacts metadata event. This should allow Toaster to
> >> > handle artifacts in simple way and remove a lot of current Toaster code doing guess work.
> >> >
> >> > To generate manifests for do_image_complete and do_populate_sdk they have been put under sstate control.
> >> >
> >> > To avoid storing big files(images and sdk installer) in sstate new variable SSTATE_SKIP_CREATION has been
> >> > set in image.bbclass and populate_sdk_base.bbclass and sstate code was modified to avoid adding files
> >> > to sstate if SSTATE_SKIP_CREATION is set.
> >>
> >> SKIP creation of what?
> > SKIP creation of SSTATE entity.
> 
> I know but of WHAT?
Of anything we don't want to put to sstate. In this patchset it's set
for do_image_complete and do_populate_sdk, so it's for images and sdk
installer.

> SSTATE_SKIP_IMAGE_CREATION would be a clearer name for me.
I'm not sure about it. I'd need one more variable
SSTATE_SKIP_SDK_INSTALLER if I go this way.

> >> Variable name seems a little vague for me. Even
> >> needing the manifest, does it makes sense to store the image and SDK
> >> on sstate?
> > It doesn't and skipping this is a purpose of introducing SSTATE_SKIP_CREATION variable.
>
> So it is always going to be skipped?
Yes it is for do_image_complete and do_populate_sdk tasks for now.

--
Regards,
Ed



More information about the Openembedded-core mailing list