Skip to main content

The FeatureReader has the option to read "Schema and Data Features", "Data Featues" and "Schema Features".

 

In an attempt to optimize a process which converts a bunch of AutoDesk dwg's I tried to first read only the schema (the layers), determine which of the layers need to be read and feed that list in a second FeatureReader as Feature Types to Read.

 

I found out I was getting slower performance, due to the FeatureReader being really slow to read the Schema Features.

  • Reading SchemaFeatures from the dwg took 120 seconds.
  • Reading Data Features from the dwg took 2 seconds.

Reading from an empty file did not result in a different performance.

 

  • Why is this so slow? Does it have anything to do with the parameters?
  • I think the FeatureReaders FME Defaults should be "Data Features" as a lot of users would benefit from the increased performance.

Hi @nielsgerrits​,

DWG files are not really structured for random access, so FME will read the entire file, regardless of the Features to Read setting. However, since FME is reading the entire file, it is reading and caching the Data features while building a Schema from those same features. So the Data read just appears to be much faster than the Schema read, but both are really occurring at the same time.


Reply