Fix deploy.placement constraints/preferences docs #507
Conversation
8543eda to
af1ad83
Compare
Signed-off-by: Paarth Shah <mail@shahpaarth.com>
af1ad83 to
cb7f738
Compare
|
Ah I see I got somewhat ninja'ed by 20ebf93 |
|
As swarm does not apply the compose specification, I wonder this is relevant to rely on swarm docs to define those attributes in the compose specification. Better keep them with a fuzzy definition IMHO |
|
Are you ok with 20ebf93 or do you feel those attributes require better documentation ? |
|
@ndeloof I'm generally okay with it, but I think we can do better. I'd have to compare/contrast my PR with it properly (it's been a little bit) As far as implementation goes, I'm wondering if docker/cli#4863 (comment) is a good example at all, or whether it's purely hypothetical at this point. And in the general case, I'd cite that the compose-spec is intended to be a superset of docker compose 2 and 3: https://docs.docker.com/compose/compose-file/14-include/
Therefore I expect everything in compose file format 3 to be de-facto transferred to the compose spec, or explicitly marked for deprecation + removal when the implementation of something better is available. I vote for the former because of the existence of PRs like the above that will provide implementation. |
What this PR does / why we need it:
This PR is to fix the docs to match
compose-spec.json's ANDdocker compose's implementation for each of:deploy.placement.constraintsdeploy.placement.preferencesThe docs have been modified to match the practical implementation of the spec as-is, including its usage for docker swarm: https://docs.docker.com/engine/swarm/services/#control-service-placement
Which issue(s) this PR fixes:
Fixes #506
Renders docker/compose#11853 irrelevant