[kepler-dev] 1st and 2nd run of the same workflow give different outputs

[Gongjing Cao]

Bertram,

Thanks for your explanation:)

On Mon, Apr 19, 2010 at 12:00 AM, Bertram Ludaescher
<ludaesch at ucdavis.edu>wrote:

> Gongjing:
>
> Just to make sure I understand what behavior you *want* to see, i.e., deem
> the "right" behavior in your case.
> But first, let me see whether I understand PortParameter right:
> I thought one use case for PortParameter is as a "hybrid" between a
> conventional (static) parameter, and a (dynamic) data port:
>
> (1) When the PortParameter is *not* connected to an upstream port, then it
> acts as a conventional parameter. In particular, parameter values are not
> "consumed"  -- they are available for actor firings as often as desired.
> (2) On the other hand, when the PortParameter *is* connected to an upstream
> port, it behaves as a "regular" port, i.e., parameter values are  "consumed"
> similar to (data) tokens on any other port. Possibly with the caveat that
> there is a default value (akin to a Delay actor).
>
> First, it would be good to confirm how close to reality (or the desired
> reality ;-) the above conceptualization (1) and (2) is.
>
> Back to the desired behavior:  Gongjing, it seems you want the
> PortParameter to ignore the default value and take all values always from
> the upstream actor even on the first run. Is that so? Wouldn't then a
> "conventional" (data) port the better choice?
>

Yes, I want PortParameter to ignore the default value and take all values
from the upstream actor even on the 1st run.

I tried to use "conventional" (data) port in workflows. I attached two
workflows:
1, dataPort-step.xml
RAMP actor already has an IO input port named "step", which is also a
parameter of this actor, so I can let this parameter/IOPort accept value
even on the 1st run.
2, dataPort-start.xml
In order to let the other parameter-"init" to work the same way, I added a
IO port-"start" from GUI, and put "start" in the configuration of "init".
But seems it is not supported in such a way. This IO Port -"start" can't
pass the value to the prarameter to "init" parameter.

Maybe we should consider to allow IO ports to pass values to all parameters?
Don't know if it will have any side effect.

Gongjing



> Bertram
>
>
> On Fri, Apr 16, 2010 at 2:23 PM, Edward A. Lee <eal at eecs.berkeley.edu>wrote:
>
>>
>> This is how PortParameter works.
>>
>> When an input arrives on the port during a run, its value
>> is stored in the corresponding parameter, overwriting the
>> previous value. Thus, by the end of the run, there is no
>> longer any record of the initial value.
>>
>> It is arguable whether it should actually work this way,
>> but the fact is, this is how it works.
>>
>> Edward
>>
>>
>>
>> On 4/16/10 10:34 AM, Gongjing Cao wrote:
>>
>>> Hi,
>>>
>>> I came to an interesting workflow when I try to use PortParameter. It
>>> generates different outputs at the 1st run and the 2nd (and after) run.
>>>
>>> Seems the PortParameter only take the default value in the 1st run,
>>> instead of taking the value which is passed from the previous actor.
>>>
>>> Does anybody know why? How can I let this workflow always give the right
>>> output as in 2nd run.
>>>
>>> Thanks!
>>>
>>> Gongjing
>>>
>>>
>>>
>>> _______________________________________________
>>> Kepler-dev mailing list
>>> Kepler-dev at kepler-project.org
>>> http://mercury.nceas.ucsb.edu/kepler/mailman/listinfo/kepler-dev
>>>
>>
>> _______________________________________________
>> Kepler-dev mailing list
>> Kepler-dev at kepler-project.org
>> http://mercury.nceas.ucsb.edu/kepler/mailman/listinfo/kepler-dev
>>
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mercury.nceas.ucsb.edu/kepler/pipermail/kepler-dev/attachments/20100419/43904cb9/attachment-0001.html>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: dataPort-step.xml
Type: text/xml
Size: 19808 bytes
Desc: not available
URL: <http://mercury.nceas.ucsb.edu/kepler/pipermail/kepler-dev/attachments/20100419/43904cb9/attachment-0002.xml>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: dataPort-start.xml
Type: text/xml
Size: 20101 bytes
Desc: not available
URL: <http://mercury.nceas.ucsb.edu/kepler/pipermail/kepler-dev/attachments/20100419/43904cb9/attachment-0003.xml>