proposal: universal id #115
Conversation
|
Added a new "back-compat" section with an overview of the effected commands and an outline for supporting the Cylc7 syntax until Cylc9. |
| [~user[/]]flow//cycle/task/job | ||
|
|
||
| # expanded form with selectors (for quick querying of objects) | ||
| [~user[/]]flow[:state]//cycle[:state]/task[:state]/job[:state] |
There was a problem hiding this comment.
[~user[/]]flow
What if flow has / in it (path hierarchy structure) .. How do you know if user has been included? is it the ~?
cycle[:state]
I thought we ruled out : colon? Because some ISO datetime uses it
There was a problem hiding this comment.
is it the ~?
Yep.
Should make it nice and explicit when working with other users workflows, ~ isn't a valid character in a workflow registration.
I thought we ruled out : colon? Because some ISO datetime uses it
I think we should be good to use the colon because Cylc doesn't use colons for its cycle point format, and can't because colons are not safe in file paths.
| > **Note:** All syntax implemented here represents functionality currently | ||
| supported in the GraphQL schema. | ||
|
|
||
| > **Note:** Currently in GraphQL `state` is a separate argument, there is no |
There was a problem hiding this comment.
Currently our GraphQL API can do both, this was intentional.. I can't recall the exact reason, but I think it was so you could do blanket exclusions and/or inclusions of states while also picking specific named tasks (?)
|
|
||
| cylc trigger OPTIONS ID | ||
| cylc trigger my_workflow//123/task | ||
| # currently: cylc stop my_workflow 123/task (or task.123) |
There was a problem hiding this comment.
Typo: should be cylc trigger not stop
| cylc trigger my_workflow//123/task | ||
| # currently: cylc stop my_workflow 123/task (or task.123) | ||
| cylc trigger my_workflow//123/*:failed | ||
| # currently: cylc stop my_workflow 123/*:failed |
| * Enables multi-user control from the CLI via the UIS without Cylc Flow having | ||
| to know anything about authorisation. |
There was a problem hiding this comment.
Not sure about this (but not important at this stage)
|
|
||
| ## Back Compat | ||
|
|
||
| We should be able to support the old legacy interface for the time being. |
There was a problem hiding this comment.
I think I'd be happy to drop the legacy interface for most commands and only worry about those we think may have been used within scripts/workflows (e.g. broadcast).
Adopt a universal identifier throughout all Cylc projects:
REGbeing an optional positional argument before required positional arguments(from the rose suite-run proposal).
Ping @dwsutherland