[oe] [doc] defining OpenEmbedded
dfarning at gmail.com
Wed Oct 24 17:36:21 CEST 2007
On Wed, Oct 24, 2007 at 07:01:30AM -0400, Cliff Brake wrote:
> On 10/24/07, David Farning <dfarning at gmail.com> wrote:
> > Getting started is coming along nicely.
> > In the second case referring to OE as a collection of metadata is
> > confusing. Metadata is data about the data. OE is actually a
> > collection of patches, task lists, and configuration files. I was
> > wondering if it would be more clear to refer to oe as either a ruleset
> > or task lists.
> The term metadistribution seems to be somewhat accepted:
> In the most basic sense, a bitbake recipe is data about data (the
> source that needs built), so I think the term fits. Just my opinion.
I like metadistribution. Metadistribution is a term referring to an
adaptable system that is capable of building itself to the user's
But, using the defination at
does not seems to apply.
Metadata is data is data which describes the data. ie
A recipe on the other hand define what ingredients to use. Source
files, patches, configs.... Then states what tasks to perform on the
ingredients to create the desired result. Binaries compiled for an arch,
added to a filesystem, bundled in an image.
Patches contain the changes necessary to make a general package work on a
Classes are a more abstract form of recipe.
Machine.conf and Distro.conf provide a high level abstraction for the
developer to define what classes and recipes to call to create a
distribution with the desired characteristics.
Am I missing anything in this very high level description?
More information about the Openembedded-devel