Integration of VEDA-FE and VEDA-BE

[AKanudia]

There is a suggestion to integrate the two VEDA applications. I need to understand the motivation behind this suggestion. Perhaps it can be fulfilled in some other way.

My views:

•  VEDA-FE is quite tied to TIMES, but VEDA-BE is not. It can be used to view any flatfile after modifying the VD file headers appropriately. VEDA-BE is actually a different application.
  


•  VEDA-BE would become unusable when any lengthy SYNC or batch run process is going on.
  

And in case the underlying desire was to see input data along with the results, integration will not help there at all. Input data (as it exists in VEDA-FE) cannot be related to results unambiguously for several reasons, vintage and case definition being the primary ones. The only clean way to achieve this would be to dump the input (in VEDA-BE format) along with the DD files, for each case. We should be able to do this quite easily with Antti’s help.

[Gary]

I believe much of the spirit/need behind this request will be addressed by the auto-importing of run results being added to the VEDA platform. It will essentially give a seamless feel between FE & BE.

[AKanudia]

Some details to help us discuss this: VEDA-BE import process has two main parts:

1. Read the text VD files into an _R database
   


2. Update the lists of unique elements for each dimension that are stored in _Y database
   

Bulk of the time is spent in the first step. I propose a separate EXE that will make an _R database out of the VD file and leave it in the same location as the VD file. When you import results in VEDA-BE, it will simply move the _R databases to the appropriate folder, if they exist, and continue with step 2.

I see a few reasons for leaving a seam in place, including:

• I could be working in VEDA-BE when the run finishes… I won’t want to be interrupted.
  


• It will be less elegant to do step 2 for each scenario separately… it should be done for a bunch of scenarios.
  

And a related suggestion for users who are running very large number of scenarios: you are likely interested in a few key results in such cases. You could set up a scenario file with non-binding UCs that create those results. With Antti’s help, we could arrange to create a “summary reports only” option that only dumps User_Con and the *M attributes. We could even think of offering more flexibility here, like collapsing the vintage dimension, as someone has suggested. This will make the VD files very small and significantly improve the experience of working with many scenarios in all aspects of VEDA-BE (not just scenario import).

[Gary]

I follow your logic in terms of perhaps at times not wanting to have VBE auto-importing results, to which end something on the option form would disable to feature as needed.

However, when doing say batch runs overnight it would be nice to have all loaded & ready to go in the morning. Having said this, if step 2 is going to be that quick (even for a large number of runs of good size models) then perhaps not that important.