Question

feature writer: "internal error" when there is no data to write, how to avoid?

  • 10 December 2019
  • 5 replies
  • 46 views

Userlevel 3
Badge +18

I have a workspace reading a polygon which is buffered and send to several featurereaders that read data from several sources (shapefiles, WFS, dwg,…).

Data is selected within this searchpolygon (using a spatial filter in the featurereaders) and the selection is written again using featurewriters.

All the summary-ports of the featurewriters are going to a featureholder (from the hub): when all data is read, written and ready a new process of map production starts based on the original polygon.

This works very well, except when there is no data to select from a certain datasource/featurereader, then I get this error:

2019-12-10 09:54:57| 1.1| 0.0|ERROR |FeatureWriter_11 (WriterFactory): Internal Error: Failed to thin feature table
2019-12-10 09:54:57| 1.1| 0.0|ERROR |Internal Error: Failed to thin feature table

 

Is this something buggy or excepted behavior? What is the best way to ‘catch’ this situation and still go on with the map production flow when all other data is written?

A simplified workspace attached that reproduces my problem (first creator gets the error, second one gets the normal workflow).

FME(R) 2019.1.3.0 (20191007 - Build 19642 - WIN64)

Thank you very much!


5 replies

Userlevel 5
Badge +25

I think there's something odd going on with the FeatureReader before it. Note that it says there's 0 features coming through the Generic port:

I tried adding a Logger to it, which shows 0 features coming in and 0 going out, but doesn't log anything. I worked around it by adding a GeometryFilter, having it filter on NULL geometry and then connect the Unfiltered port to your FeatureWriter, that makes it all work properly (i.e. FeatureWriter not throwing up an error)

Whatever is going on with the top FeatureReader is not happening in the bottom one, that one doesn't show any output either but does not trigger an error in the FeatureWriter following it.

I think Safe's support team would be very interested in seeing this and I would recommend you send it in. For the record, I'm seeing this behaviour in build 19643 on Mac.

 

Userlevel 3
Badge +18

I think there's something odd going on with the FeatureReader before it. Note that it says there's 0 features coming through the Generic port:

I tried adding a Logger to it, which shows 0 features coming in and 0 going out, but doesn't log anything. I worked around it by adding a GeometryFilter, having it filter on NULL geometry and then connect the Unfiltered port to your FeatureWriter, that makes it all work properly (i.e. FeatureWriter not throwing up an error)

Whatever is going on with the top FeatureReader is not happening in the bottom one, that one doesn't show any output either but does not trigger an error in the FeatureWriter following it.

I think Safe's support team would be very interested in seeing this and I would recommend you send it in. For the record, I'm seeing this behaviour in build 19643 on Mac.

 

thank you @redgeographics, i've also added my buildinfo to the original question

I thought the second reader never started reading since the process stopped at the first reader, but as you say the behavior of both is just different. So I started digging a little deeper and thought I was going crazy when the error continued to occur even when all settings and even datasources were equal. Until I noticed the "summary annotation" of both looked slightly different:

Seems I missed the fact that the top reader is set to Esri Shapefile (Tech Preview) and the bottom one to the "oldskool format" Esri Shapefile (*). If I change the tech preview to the regular format, the error does not occur!

Strange and seems to be a bug, as you suggest I will send this to safe-support.

Thank you for your time again!

(*) I am not entirely sure, but I think the format was both on <Guess from Dataset> when I added the readers. Anyway, if I now add a new FReader to the workspace, select a shapefile, it will set the format to tech preview by default.

Userlevel 3
Badge +18

filed as FMEENGINE-62465 at safe

specific to the newer SHAPEFILE reader and the 'Bounding Boxes...' spatial filter in the FeatureReader

solution --> don't use the 'tech preview' shapefile in these cases...

Userlevel 3
Badge +18

This problem has now been fixed in the latest FME 2020 betas .

Userlevel 3
Badge +18

filed as FMEENGINE-62465 at safe

specific to the newer SHAPEFILE reader and the 'Bounding Boxes...' spatial filter in the FeatureReader

solution --> don't use the 'tech preview' shapefile in these cases...

This problem has now been fixed in the latest FME 2020 betas .

Reply