[Openembedded-architecture] RFC: RM_OLD_IMAGE as default behaviour

[Martin Jansa]

On Wed, Oct 12, 2016 at 03:45:45PM +0200, Philip Balister wrote:
> I don't care. Someone did ask how to remove old images during the BoF,
> so at least one user would prefer the be default.

Same here.

Where we care about the build artifacts (most our jenkins builds), then
we move them after the build to somewhere else already.

For local incremental builds it's already hard to match DATETIME with
the metadata used to create the image, so it will encourage people to
move them somewhere else with some additional name to distinguish them.

But I've noticed one strange behavior that qemuboot.conf file seems
being kept in incremental images, so I have multiple copies of qemuboot.conf
and only one image next to it - I'll double check if it's only my environment
or the same happens with default nodistro build as well.

core-image-minimal-qemux86.hdddirect
core-image-minimal-qemux86-master-20161012105326.qemuboot.conf
core-image-minimal-qemux86-master-20161012111034.qemuboot.conf
core-image-minimal-qemux86-master-20161012120038.qemuboot.conf
core-image-minimal-qemux86-master-20161012120336.qemuboot.conf
core-image-minimal-qemux86-master-20161012134817.qemuboot.conf
core-image-minimal-qemux86-master-20161012144327.ext4
core-image-minimal-qemux86-master-20161012144327.hdddirect
core-image-minimal-qemux86-master-20161012144327.manifest
core-image-minimal-qemux86-master-20161012144327.qemuboot.conf
core-image-minimal-qemux86-master-20161012144327.tar.gz
core-image-minimal-qemux86-master-20161012144327.vmdk
core-image-minimal-qemux86.qemuboot.conf
core-image-minimal-qemux86.rootfs.ext4
core-image-minimal-qemux86.rootfs.manifest
core-image-minimal-qemux86.rootfs.tar.gz
core-image-minimal-qemux86.vmdk

> On 10/12/2016 03:21 PM, Joshua Lock wrote:
> > tl;dr: RM_OLD_IMAGE behaviour is effectively enabled by default in
> > master/morty — any objections?
> > 
> > Around a month ago a change landed in OE-Core to put image deployment
> > under sstate control[1]. 
> > 
> > This had a side-effect of removing old images from DEPLOY_DIR_IMAGE
> > when a new image is built[2], effectively making RM_OLD_IMAGE[3] the
> > default behaviour and making that variable redundant.
> > 
> > Several of us run with RM_OLD_IMAGE and didn't notice the change in
> > behaviour, others noticed and assumed it was an intended side-effect of
> > the change. 
> > 
> > This is arguably a significant behaviour change so I wanted to ensure
> > it was more visible before we tag a release with this behaviour.
> > 
> > Do folks have strong opinions on whether we should revert to the
> > previous behaviour of not removing images (also sdk's and toolchains)
> > when deploying a new version, or whether we should keep the new
> > behaviour, document it and remove the RM_OLD_IMAGE variable? Or perhaps
> > something else entirely?
> > 
> > Thanks,
> > 
> > Joshua
> > 
> > 1. http://git.openembedded.org/openembedded-core/commit/?id=d54339d4b1a
> > 7e884de636f6325ca60409ebd95ff
> > 2. https://bugzilla.yoctoproject.org/show_bug.cgi?id=10419
> > 3. http://www.yoctoproject.org/docs/2.1/ref-manual/ref-manual.html#var-
> > RM_OLD_IMAGE
> > _______________________________________________
> > Openembedded-architecture mailing list
> > Openembedded-architecture at lists.openembedded.org
> > http://lists.openembedded.org/mailman/listinfo/openembedded-architecture
> > 
> _______________________________________________
> Openembedded-architecture mailing list
> Openembedded-architecture at lists.openembedded.org
> http://lists.openembedded.org/mailman/listinfo/openembedded-architecture

-- 
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: 163 bytes
Desc: Digital signature
URL: <http://lists.openembedded.org/pipermail/openembedded-architecture/attachments/20161012/dcf7d3a6/attachment.sig>