I have a project and it seems that some of my end readers are reading in files before the workbench gets to them. It would not be a issue but its sending data through a feature merger before it should. Is their anyway to force the workbench to start at one area and then not go to the next transformer until the previous one is completely done? Some sort of hold. I know this would be less efficiant but may solve my issue.
Page 1 / 1
There is a FeatureHolder transformer available.
And maybe the option Supliers first on the merger is usefull?
The order in which FME will read the Source datasets is the order in which the sources are shown in the Navigator. Move a reader dataset up in the Navigator and it will be read earlier.
There is a FeatureHolder transformer available.
And maybe the option Supliers first on the merger is usefull?
Thank you, that is exactly what i was looking for.
Yes, FME will start reading as much data as possible, and processing it simultaneously because it's more efficient that way.
In general, transformers like the FeatureMerger are what we call Group-Based. They wait until all processing up to that point is done before starting.
Other transformers are what we call Feature-Based. They process data immediately, but it's not an issue because the transformation of one feature doesn't impact another.
The only place it is usually an issue is when we start writing to a database that has a unique constraint (and the part of your workspace that was supposed to delete a record hasn't taken place yet), or the order of table writing is important for joins.
In short, it should never really be a problem. The FeatureHolder does stop things, but the occasions you need it are really, really rare.
So, although you've flagged this as resolved, can you still attach the workspace or include a screenshot? I'd like to know what it is that is running before you think it should, and how it is causing a problem.