I think what you're doing is basically correct but I would offer the following suggestions / comments:

- The FeatureWriter is guaranteed to not output any features on the Summary port until all features have been written, so your FeatureHolder+Sampler, following the FeatureWriter are not necessary

- From 2018 onwards, the FeatureWriter can be configured to allow all input features to also be output (i.e. enable the "One per Feature Type" Output Port option. If you do this, you shouldn't need a separate branch for your statistics calculation

- You can also get a record count for each written feature type from the FeatureWriter by making use of the "Feature Type List Attribute" Summary Attribute, so your StatisticsCalculator should not be necessary

If you use all those tips I think you should be able to reduce the complexity of your workspace considerably....

@agelfert

We stage the processes. After verification, move to next stage etc. (to prevent wrting incomplete succeses etc.)

Run the script and then parse the log file from the workbench.

I use regexp to extract relevant information. Then test it.

Here are some regexp of a old workspace I have:

((\\d-]+)\\s+((\\d\\:]+)((\\d\\|\\.\\s]+)(WARN)\\s+\\|((\\++)|(Attribute.*`(.*)'.*value\\s+`(.+)'))

which reads warnings from the log. (this is from a 2016 workspace)

You can use same technique to get all the stats.

Create report and send email/twitter or wahthavewegotmore.

-------------

You can also use bat files (though stringparsing is less powerfull), here is a sample part:

fme.exe "%Scripts%\\ARIS\\ARIS meldingen_v3 EnvPara.fmw"

FIND "Translation was SUCCESSFUL" "%Scripts%\\ARIS\\ARIS meldingen_v3 EnvPara.log" >nul

Great feedback. Thanks all! I have a whole bunch of workspaces to "fortify" with this kind of reporting. So I will try and incorporate your suggestions.