Skip to main content
Question

FeatureWriter doesn't overwrite every file in ZIP folder

  • April 1, 2026
  • 2 replies
  • 33 views

antja
Contributor
Forum|alt.badge.img+11

Hi everyone,

I’m currently facing an issue with CSV FeatureWriter transformer writing in a ZIP folder. We recently migrated from FME Workbench 2020 to 2025 and the behaviour of the writer changed in between.

  1. 2020: when writing a CSV in a ZIP folder, the entire ZIP folder would be overwritten. Meaning that if it contained several files, only the newly written CSV would remain.
  2. 2025: when writing a CSV in a ZIP folder, only the specified CSV file in the folder is overwritten. Meaning that all the other files are still existing. I get the same result with FME2024.1.3.

Main question here: Which FME version has the correct behaviour ? Case 1 or case 2?

If it’s case 2, then it means that I will have to update all the workspaces in my infrastructure that are writing files in a ZIP folder, leading to a lot of extra work → I probably need to clean up the ZIP folder upstream through a SystemCaller or PythonCaller. Have you other options in mind?

Please find enclosed a folder with relevant resources:

  • sample CSV
  • FME workspaces (2020 and 2025)
  • ZIP folder ‘TEST_CSV.zip’ with 2 files in it, when using FME2020 one file gets overwritten and the other deleted

2 replies

hkingsbury
Celebrity
Forum|alt.badge.img+70
  • Celebrity
  • April 2, 2026

Main question here: Which FME version has the correct behaviour ? Case 1 or case 2?”

Great question! And I can see two both sides….. It really depends on whether you see a .zip as a folder/directory, or a file…..

I’d be more on the side of 2025s approach, as FME just looks at a .zip as a directory and opens it on the fly….

But I can fully understand why you’d expect the 2020 approach as a .zip is a file* (as far as your operating system is concered)

*or a container, but at that point, is it just semantics? as you could call any file a container for various content types


redgeographics
Celebrity
Forum|alt.badge.img+62

I agree with ​@hkingsbury in saying that the FME 2025 behaviour feels like it’s the correct one.