Fix output consistency check when there are offsets. #6526
Conversation
hjoliver
force-pushed
the
fix-graph-offset-opt
branch
3 times, most recently
from
0cb877a to
811699f
Compare
| # Don't check the left hand side if this has a non-terminal RHS: | ||
| param((('a &', 'b[-P42M]'), {}), None), | ||
| # Check the left hand side if this has a non-terminal RHS: | ||
| param((('a &', 'b[-P42M]'), {}), 'Null task name in graph'), |
There was a problem hiding this comment.
a & => b[-P25] => b
On master this validates OK and the test is confirming that? 🤯
There was a problem hiding this comment.
Fixed against master with this change hjoliver#63: I've failed to come up with other bad scenarios, but this is bad enough.
hjoliver
force-pushed
the
fix-graph-offset-opt
branch
from
811699f to
691f6f5
Compare
Co-authored-by: Tim Pillinger <[email protected]>
hjoliver
force-pushed
the
fix-graph-offset-opt
branch
from
691f6f5 to
b0ec142
Compare
| ┆ ( | ||
| ✓ ┆ started | ||
| ⨯ ┆ and succeeded | ||
| ┆ ) |
There was a problem hiding this comment.
This is now consistent with the completion condition if no offset is present.
hjoliver
force-pushed
the
fix-graph-offset-opt
branch
from
f3592ef to
697f923
Compare
| # The word "optional" is common to both. Not ideal! | ||
| 'optional', | ||
| id='succeed-or-failed-implies-success-optional' | ||
| ), |
There was a problem hiding this comment.
Took your test cases with attribution (via commit message) @wxtim
| @@ -102,8 +102,8 @@ async def test_over_bracketed(flow, scheduler, reftest): | |||
| 'final cycle point': '2013-12-25T12:00Z', | |||
| 'graph': { | |||
| 'T12': ''' | |||
| (a[-P1D]:fail | b[-P1D]:fail | c[-P1D]:fail) => d | |||
| a & b & c # Implied by implicit cycling now... | |||
| (a[-P1D]:fail? | b[-P1D]:fail? | c[-P1D]:fail?) => d | |||
There was a problem hiding this comment.
This is absolutely reasobable in the case of recurring cycling, but is it reasonable in general?
i.e. I think that this should fail validation
[graph]
R1/1 = a
R1/2 = a[-P1]:failbut I'm not sure that this should, because a at -P1 is not the same task as a.
R1 = a[-P1]:fail => aI think I may be happy to gloss over this, because the latter recurrance isn't very meaningful.
| suicide = (suicide_char == self.__class__.SUICIDE) | ||
| optional = (opt_char == self.__class__.OPTIONAL) | ||
| if output: | ||
| output = output.strip(self.__class__.QUALIFIER) | ||
|
|
||
| if name in self.family_map: | ||
| fam = True | ||
| mems = self.family_map[name] | ||
| rhs_mems = self.family_map[name] |
There was a problem hiding this comment.
rhs_members for easier reading please?
| ), | ||
| param( | ||
| "foo[-P1]:succeeded? | foo[-P1]:failed? => foo", | ||
| # Order of processing can result in different error messages |
There was a problem hiding this comment.
I fixed this in my version of the pr hjoliver#62
Close #6523
Check List
CONTRIBUTING.mdand added my name as a Code Contributor.setup.cfg(andconda-environment.ymlif present).?.?.xbranch.