Re: [SED-ML-discuss] Fwd: [kisao:feature-requests] #31 Number of Runs

[Lucian S. <luc...@gm...>]

Just to summarize the proposals I've seen so far, there are at least four
possibilities on the table for repeating SED-ML runs:

1) Introduce a new KiSAO term, and use it like any other KiSAO parameter.
2) Introduce a new attribute to the SED-ML 'Task' class that would tell you
how many times to repeat the Task.
3) Introduce a new attribute to the SED-ML 'RepeatedTask' class that would
be a 'simpler' way of repeating tasks.
4) Just use the 'RepeatedTask' class as-is to repeat tasks.

Chris is a staunch advocate of approach 1 and/or 2.

I am not convinced we need anything other than option 4.

Both of us agree that it would be best if there was exactly one way to
encode this information: neither of us want one group of people using
approach 1 while a different group of people use approach 4, for example.

If KiSAO did indeed add the term, I would prefer SED-ML users use that and
not option 2.


The seed and the resetting thereof is a completely different issue, which
needs to be addressed in SED-ML, but as noted, the term is indeed already
in KiSAO; we just need to clarify its semantics when we use it:
https://github.com/SED-ML/sed-ml/issues/30


(By the way--I've the new google groups list to this email's cc list, but
did not remove the sourceforge list, in the vain hope that perhaps some day
this discussion will be readable on one archive or the other.)

-Lucian