OpenBMC - how best to do meta-hpe commit

[Andrew Jeffery]

Hi Mike,

On Sat, 20 Jun 2020, at 01:07, Garrett, Mike (HPE Server Firmware) wrote:
> Hi Brad and all,
> 
> Thank you for creating meta-hpe.  I'd like your opinion on the best way 
> to do our initial meta-hpe commit.  Its currently 168 files split into 
> (1) a machine layer for our DL360, (2) a BMC SOC layer, (3) a common 
> layer for elements common to HPE platforms.  I know the team prefers 
> small commits but would like your thoughts on this since it is more or 
> less atomic.

Right, smaller commits are preferable.

It does look like you already found some boundaries along which it might be 
possible to split the work: your machine/soc/common distinction above. However, 
my guess at the best course of action is to push the one big commit you have to 
gerrit, which at least allows interested people to get their eyes on the code. 
>From there we can figure out how to divide the problem up if necessary (for 
example, if you're carrying kernel patches, how we can get them into the 
openbmc kernel tree rather than apply them via bitbake). Just be prepared to 
iterate until we can digest the changes.

> 
> I assume this would get committed to 
> https://gerrit.openbmc-project.xyz/openbmc/meta-hpe and go through the 
> normal Gerrit review process right? 

Yes.

> I've looked and not been able to 
> find good examples of other "1st commits" to see how it's been done in 
> the past.
> 

Yeah, we've had similar questions come up in the past and I'm not sure there 
have been any good answers. The key thing is to try to get all the pieces in 
the right places; get u-boot patches into our u-boot fork, get kernel patches 
into our kernel fork etc. Maybe we need to add an "Upstreaming your changes" 
section to 
https://github.com/openbmc/docs/blob/master/development/add-new-system.md

Hope that helps,

Andrew