Hi,
Is it possible to build reasonably generic graphs to perform certain tasks and then “call” them from other graphs?
The intention is to reuse the graph code and not write in in each graph. If I find a bug in that particular code, I then change it in one place and not in each of the individual graphs.
Thanks
Des
Hello Des,
I think you can use the RunGraph component. Take a look at the following documentation:
http://www.cloveretl.com/_upload/clover … l#d0e22042
Best regards
Tomas Waller
Thanks for the reply. RunGraph is fine, but I still need to pass the stream into the new graph.
this means if I have a 30 odd fields in the first graph, I somehow need to get these 30 fields to the next graph. I have done it now by writing the data to a temp dB table and then reading this same dB table on the next graph when called, but wondering if there is a better way?
Thanks
Des
This type of data flow handover is currently impossible. Your solution with temp files is correct. I have never tested it before, but you can consider to use named pipes under unix-like systems for better performance. The most fastest way to pass some date between two graphs is probably via our proprietary format, which is handled by Clover Data Reader/Writer.
Regarding your needs, exactly this should be available in Clover 3 (http://wiki.cloveretl.com/doku.php?id=roadmap:clover3) and this will be called ‘Sub-Graphs’.