pkglistgen.py is a self contained script to generate and update OBS products for openSUSE and SLE. It works on the products and its staging projects and ports.
The main input is a package named 000package-groups and it will update the content of other packages from that. For that it will read YAML input from e.g. 000package-groups/groups.yml and generate .group files into 000product. The rest of 000package-groups is copied into 000product as well and it runs the OBS product converter service (See OBS Documentation for details) The generated release spec files are split into 000release-packages to avoid needless rebuilds.
The package list generator reads several files. The most important are group*.yml (traditionally only groups.yml) within 000package-groups.
The file lists the packages and their support level. It’s only necessary to list packages here that have a different level than the default level specificied in the groups. The format is plain text:
The file is a list of package lists and the special hash ‘OUTPUT’. OUTPUT contains an entry for every group file that needs to be written out. The group name of it needs to exist as package list as well. OUTPUT also contains flags for the groups.
We currently support:
default-support Sets the support level in case there is no explicitly entry in supportstatus.txt, defaults to ‘unsupported’
recommends If the solver should take recommends into account when solving the package list, defaults to false.
includes Adds package lists to the group to be solved. Allows to organize different topics into the same group. By default there are no package lists added - the package list with the group name is always there.
excludes Removes all packages from the solved groups listed. Used to build addons to main products.
conflicts Sets package groups not to be part of the same product. Influences the overlap calculation only.
Be aware that group names must not contain a ‘-‘.
You can also adapt the solving on a package level by putting a hash into the package list. Normally the package name is a string, in case it’s a hash the key needs to be the package name and the value is a list of following modifiers:
Note that you can write yaml lists in 2 ways. You can put the modifier lists as multiple lines starting with -, but it’s recommended to put them as [M1,M2] behind the package name. See the difference between pkg4 and pkg5 in the example.
OUTPUT:
- group1:
includes:
- list1
- list2
- group2:
default-support: l3
recommends: true
includes:
- list3
excludes:
- group1
conflicts:
- group3
- group3:
includes:
- list2
group1:
- pkg1
group2:
- pkg2: [locked]
- pkg3
group3:
- pkg4: [x86_64]
list1:
- pkg5:
- x86_64
list2:
- pkg6: [recommended]
TODO
If 000package-groups contains a file named summary-staging.txt, the bot will trigger a diff mode on staging projects. It will create an equal summary-staging.txt in 000product and create a comment with a human readable diff in the staging project. This comment can be replied to. If the reply starts with ‘approve’, staging accept will apply the diff to this txt file. If the reply starts with ‘ignore’, the bot will continue with the pipeline and do nothing on staging accept.
This way simple changes to the summary can be accepted without a submit request (of which there can only be one at a time).