Question

Change in the behavior of the Bufferer transformer between FME 2017 and FME 2020


Hello all,

I am facing a change in the behavior of the transformer Bufferer. I created a script in 2017 using Bufferer to generate eroded polygons with the same number of points as the original polygons. When I wanted to use this script with FME 2020, I found that the eroded polygons had more points, which disrupted the rest of the processing chain.

 

Is there a way to recover the previous behavior?

 

Here are some pictures to illustrate the problem I am facing.

 


7 replies

Userlevel 6
Badge +31

You can install fme 2017, create a workbench with a bufferer and save it. Open the workspace with fme 2020 and don't upgrade the transformer, now you can copy paste it to the workspace you need.

This is what I did, but it doesn't work, the behavior changes anyway. 😕

The script I shared was created with FME 2017, but if you reload it with FME 2020 and run it again the results are quite different.

You can install fme 2017, create a workbench with a bufferer and save it. Open the workspace with fme 2020 and don't upgrade the transformer, now you can copy paste it to the workspace you need.

This is what I did, but it doesn't work, the behavior changes anyway. 😕

The script I shared was created with FME 2017, but if you reload it with FME 2020 and run it again the results are quite different.

Userlevel 1
Badge +11

Hi @lavairye​ Can you attach your original workspace and some sample data here? thanks

Hello @danatsafe​, Attached is the original workspace (2017) and the same one recorded with FME2020. I add a data set to use with these workspaces.

Userlevel 1
Badge +11

Hi @lavairye​ The change was intentional starting with FME 2020.0 to fix some other issues with the Bufferer. The behaviour you are seeing in FME 2017 is related to the Miter distance which wouldn't have the same result for every vertex. A Generalizer with a very low tolerance in FME 2020+ removes the extra vertices. Please see the attached workspace, first created with FME 2017 and then edited with FME 2020.2, Build 20825.

Hi @lavairye​ The change was intentional starting with FME 2020.0 to fix some other issues with the Bufferer. The behaviour you are seeing in FME 2017 is related to the Miter distance which wouldn't have the same result for every vertex. A Generalizer with a very low tolerance in FME 2020+ removes the extra vertices. Please see the attached workspace, first created with FME 2017 and then edited with FME 2020.2, Build 20825.

Thanks @danatsafe​, I suspected that the generation of additional points where they are located must be related to the Miter. I will implement your method and test on my full datasets.

Reply