Refactor support for distro/version pattern rules in makefiles #2960
Conversation
webbnh
added
Code Infrastructure
packaging
Huh; I just noticed this comment while flipping through tabs. It is possible to have multiple PRs pointing at a single Jira issue, and they all (in theory) get linked. This can be awkward if the issue spans sprints, since there's no easy way to split this so the completed work gets credited to the sprint. Another option is to create sub-tasks under the story and connect each PR to its own sub-task. That's more or less the "official / intended" way to do it; but that can also be awkward when the story doesn't complete in a sprint because Jira makes splitting stories and moving incomplete sub-tasks much, much harder than it ought to be. (Sigh.) A third option is to create another story under the same epic. This doesn't have either of those disadvantages, but of course it also weakens the Jira connectivity between the two, especially if the epic is already rather large. (You can improve that a bit by creating dependency links between the stories; but while that's easy to do it's not visible in the usual collection views and therefore of limited real use.) |
|
Ready to go except for a |
Create a new utility file, util.mk, which holds handy widgets for implementing makefiles, and rehome the code which underpins distro/version-based pattern rules there. Also, add a couple of other utility definitions.
As I implement support for building things on/for multiple distribution/version combinations, I find myself reusing the same (or similar) code idioms which drive Make pattern rules where the pattern consists of a tag with the distro name and the major version (like
rhel-9,centos-8, orfedora-35). There's a modest set of boilerplate definitions which support this usage, and, rather than continue to replicate it where it's needed, this PR moves it to a common, utility makefile which can be included by the makefiles which need the definition.Also, in an upcoming PR, I will need some other standard Make magic, so I've added the definitions which support that (
_emptyand_space) to this file, as well.PBENCH-720