Give me an error instead of half writing tables!

I’ve been hit by this one too (combined with an attribute name that seemed to imply it was a numerical field… “numberoftracks”…. turned out the possible values were “one”, “two” and “lots”)

Yup. I‘ve been bitten by this one too. In general you’re never 100% sure if stuff was (partly) written or not, when you get these kinds of warnings. So then the state of your data becomes completely unclear, and it takes time to check the integrity afterwards…

@redgeographics : How about “diameter” on a pipeline? You’d think that be a numerical value. Some of the actual values I found: “2.5 inch”, “2,5 inch”, “20 centimeter”, “45”, “twee decimeter”….. 

But I must admit that “lots” made me laugh 😄

Oooof, that is just nasty…

I must admit I did exaggerate for comedy effect, the actual values were “Enkelspoor”, “Dubbelspoor” and “Meervoudig”. End result was that my database with Top10NL data had no railways...

Try to extrude a pipeline into a 3D object by a value of “1.2 feet”….

“End result was that my database with Top10NL data had no railways”

And that is what gets me. If you do not carefully dig through the logfile, you’ll miss the fact that some of the data wasn’t actually written, when the translation log says that it was succesful… Been working with FME for a good 12 years or so now, and this still trips me up now and then...

I learned to be triggered by differences in featurecounts because of this.