Re: is there a doc section on yocto distribution creation/versioning?

Bryan Evenson


-----Original Message-----
From: yocto-bounces@... [mailto:yocto-
bounces@...] On Behalf Of Robert P. J. Day
Sent: Thursday, June 25, 2015 6:29 AM
To: Yocto discussion list
Subject: [yocto] is there a doc section on yocto distribution

is there a recommended workflow for tagging/versioning one's own builds
from yocto? that is, given the collection of layers that might go into one's
local builds, in order to properly version your builds for release, one would
have to keep track of not only the checkouts of the various layers, but any
local fixes added to them, combined with all of the local conf settings used
for that build, and so on.

is there a writeup on this?
If there is a writeup on this I have not seen it. I've developed a workflow on my own for our setup that works okay. Whether it'd be a recommended workflow or not is a different question.

We have a local Git server which houses our clones of the yocto/oe layers and our own layers. We are on the dizzy branch now, so we have a local dizzy branch on our Git servers that are a clone of the official repos with a few extra commits (I believe there were a few master patches we incorporated to fix some issues we were seeing). We then have two layers of our own. One layer is a public layer and the other is a private layer. The public layer contains our base image recipe and all the bbappends for yocto/oe recipes to setup the image the way we want. The private layer has our proprietary applications and then a bbappend for the image to add the proprietary applications. This way we can do a build without our private layer to ensure we aren't accidentally deploying any proprietary code for FOSS compliance. We have an autobuilder setup to build the public image and the full image to verify both builds are working.

We also have a distribution recipe similar to the angstrom-version recipe that Angstrom uses for its distribution versioning. This is how our system can tell which distribution version it is on and report it to the user.

When we do a new distribution release, we tag all the layers with the distribution version name and commit the tags to our local Git server. Then we can retrieve all our local layers based up on a tag and get back to a specific release if we need to.

If anyone else has a different method that they think works better, I'd love to hear it.




Robert P. J. Day Ottawa, Ontario, CANADA

yocto mailing list

Join to automatically receive all group messages.