[Openembedded-architecture] Notes from the OEDAM "Naming Table"

Trevor Woerner twoerner at gmail.com
Fri Sep 9 14:55:28 UTC 2016


On Fri 2016-09-09 @ 03:08:28 PM, Koen Kooi wrote:
> 
> > Op 8 sep. 2016, om 22:14 heeft Trevor Woerner <twoerner at gmail.com> het volgende geschreven:
> 
> [..]
> 
> > The poky repository is vitally important because the community has decided
> > that the contents of this repository forms the baseline for the metrics of
> > this project.
> 
> I *strongly* object to the use of ‘community’ in the above. 

Oh good point, very true. That was a rather careless choice of a word. In fact
the opposite would be more correct: "a (small, but highly influential) group
has decided the poky repository is very important" ?

I can understand why the poky repository is considered important and unlikely
to go away, my point is it confuses people. It would be silly to ask that it
be abolished, so I'm asking people to consider the possibility it could be
relegated to a place outside the public's view? It can still be publicly
available, just not the repository around which The Yocto Project revolves?

This might naturally lead some to counter with: having the poky repository
makes it easier to describe bugs and recreate builds from developer to
developer to CI to etc. It's very easy to say: "I just pulled the poky
repository and noticed that the qemuarm build failed" rather than "I just
pulled <this> distro with <that> BSP plus OE/meta and my build failed for
<some> machine".

True, no argument. Which is why it would be great to have a mechanism for
fully describing builds in a completely unambiguous way (something I've been
trying to get people excited about for a while). If I could have my build
generate this one file, then I could put the contents of that file in a bug
report (or attach it), someone could grab that file, "bitbake" it, and
recreate my exact build. Currently I would have to list the contents of
multiple configuration files and still not have all the information a bug
report would need since nowhere do we record from where a repository is
cloned.

In other words, it would appear (and this is just my opinion, I'm not trying
to make anyone angry or be a troll) that in creating the poky repository
an important problem was masked? Namely that we're using too many config files
and that a build depends too much on things that happen outside of the build's
control?



More information about the Openembedded-architecture mailing list