TestFilter Version 3 Dropping Values on Tested Attributes?

Found another tidbit on this. The TestFilter is immediately downstream from a CommonLocalReprojector transformer. If I update the TestFilter but not the CommonLocalReprojector the bug seems to disappear. Maybe this is a problem with the Version 3 CommonLocalReprojector?

Sounds like the issue is with the CommonLocalReprojector. Deleting and re-adding it may work, if it was created in a workspace before ~2023, run now in 2025, even if it’s the same transformer version. If its behaviour of dropping the fme_feature_type changes if you have less than 1000 features, then report it to Safe as a bug. 

Hi @pmcpherson,

I wasn’t able to reproduce this issue when upgrading a sample workspace from 2022.2 (version 2 of TestFilter & CommonLocalReprojector) to 2025 (Build 25815). If the ctredinnick’s suggestions didn’t resolve it, could you please share a minimum workspace along with sample datasets so we can reproduce the issue and dig in further?

Also if you’re only filtering fme_feature_type with the TestFilter, I’m curious if you observe the same thing if you swapped it with the dedicated FeatureTypeFilter transformer as well.

Hi ​@pmcpherson,

I was looking around on the forum for answers because I have the same issue: the Testfilter is setting fme_feature_type to missing(null in your case). Not for all rows, just “some” of them. 

After some testing I have found two solutions:

• First you could precede the step with AttributeEncoder and set it all to UTF-8. It worked for me then.
• However I would try to remove fme_feature_type from the testfilter. In my case I had testing done on two different fields. One of which was fme_feature_type. After removing fme_feature type from the TestFilter and adding another TestFilter with the tests on fme_feature_type it worked! There were no more missing fme_feature_type fields.

​@jamatsafe It seems the TestFilter has some unintended behaviour when integrating fme_feature_type as a test parameter.
All testing was done in FME(R) 2025.1.2.0 (20250829 - Build 25630 - WIN64)

All transformers are updated to their latest version
 

The test condintions above cause the unintended behaviour. Removing the fme_feature_type else if row fixes the issue. It doesn't matter where in the else if order the row is. The answer is the same. 

I'll try to create simple workspace that highlights the problem