When I copied some features from a backup table that was created before the preceding process into that output table, I see that some nodes are not snapped anymore, just like if nodes were moved very slightly.
I suspect the reprojection could maybe round the coordinates. Is it something someone have already experienced?
If the original data has fixed length coordinates(how can it else..lol) , say 6 or 9 decimals after comma) reprojector might indeed have introduced (very slight) translation
as fme internaly does not limit itself to said fixed decimals. (more to machine limits)
It is highly improbalbe that all or any coordinates are complient to any numbertheoretic formal system...
Actually any matrixtransformation might introduce such.
You can test it with an arc, famous for such occurences.
Create an arc using the transformer, add a random angle.
Coerce to lines and then try to reattach them..mostly it will refuse latter.
You can increase the snapping tolerance of course.
depending on the formula used for the reprojection your coordinates could certainly shift or drift somewhat. If you're only reprojecting to calculate areas, you might be better off by doing something like this:
David