Wu Da-Yi10/23/2023, 4:01 AM
then need to run b first, then a. even in python code a is defined before than b. Everything works well with dfs/topological sort, sorting the node, execute the node one-by-one, until i deal with BranchNode. Normally, task or other nodes can be excuted by
. However, BranchNode seems not follow this rules. The root cause is that the
is already a
that make it really hard to call in python. I am trying to solve this problem by adding a
and add the python callable object
as a new attribute into BranchNode, I dont know if this is a correct direction to work on. Do anyone know about this?
Wu Da-Yi10/23/2023, 4:47 AM
visit all branches using
function, and then write itself into branchnode using
. So actually, all the state are written in Branchnode. The whole problem just branchnode have no good interface for python execution. (
, on the other hand, is good for python execution ). Not sure if it is a good solution just write conditional into the branchnode attribute)
Wu Da-Yi10/24/2023, 9:30 AM
in branchnode. When I am trying to execute the
cases using pyflyte remote run. Remote also not shwoing the err meesage I type in python.
if we sent my_input = 11.0 . there is no correct message shown in flyte remote.
@workflow def wf1(my_input: float): return conditional("fractions") .if_((my_input > 0.1) & (my_input < 1.0)) .then(double(n=my_input)) .else_() .fail("The input must be between 0 and 10") )
("The input must between 0 and 10")
is shown in flyte remote instead
[MalformedBranchUserError] No branch satisfied
Kevin Su10/27/2023, 7:20 AM