On 01/14/2016 07:35 AM, Nick Coghlan wrote:

On Thu, Jan 14, 2016 at 3:33 AM, Honza Horak hhorak@redhat.com wrote:

...

#topic review of how development of new scls is working rought proposal: building packages -> candidate -> sanity testing -> testing -> [wait till rhscl is released] -> tag released -> sign&release

• usual tagging process: tag -candidate -> [sanity testing] -> tag

-testing -> [better testing] -> tag -released -> sign&release

• how it will be done for collections that gonna be released in RHSCL in

the future?

• idea: such collections should stay in -testing until they are released

in RHSCL -- bad idea? shout now.

What's the scope of discussion here in terms of "sclo-*" SCLs and "rh-*" SCLs?

For "sclo-*" SCLs, I'd expect the RHSCL release cycle to be entirely irrelevant.

Yes, they may be theoretically influence in cases the sclo-* SCLs depend on some rh-* SCLs -- in that case their release is depended on RHSCL.

For "rh-*" SCLs, having them remain in -testing until the downstream release (so they become generally available to RHEL & CentOS users at the same time) seems reasonable.

So if someone wanted to take an existing rh-* package and turn it into an sclo-* package that could be updated independently of the RHSCL release cycle, they'd be free to do that.

Yes, in case someone would like to include different set of packages or deliver different version, then it effectively become a new SCL, that needs to be treated that way.

Honza