-<p>No normalization is done beforehand, so the results for the same quads may vary based on the Dataset implementation.</p>
+<p>No prior normalization is required, therefore the results for the same quads may vary depending on the Dataset implementation.</p>
Two reasons for this PR:
No normalization is done beforehand can be interpreted as: implementers should not normalize in .toString. I think the point is that normalization is not required, not that it should not be done.
No normalization […], so the results may vary could mean that the goal of not normalizing is to return different results. This change makes it clearer that the varying nature of the result is more of a consequence of not normalizing than the reason not to normalize.
Two reasons for this PR:
.toString
. I think the point is that normalization is not required, not that it should not be done.https://raw.githack.com/rdfjs/dataset-spec/e91c864c9b75d2c64fc4ef8fcea9bd9314e0d994/index.html