Roy T. Fielding wrote:
-1
This is wrong on multiple counts. First, the job of a parser is to
parse, not invent. The patch makes round-tripping of content impossible.
Second, there is no guarantee that :// implies a hostname or that a
single "/" following that hostname can be omitted -- those are both
scheme-dependent parsing rules.
Well, the patch was written based on the assumption that the comment is
correct...
The API comments should be fixed for PR 44761. Subversion should
be normalizing the URI before it is used.
... but that doesn't seem to be the case. Can this be fixed?
thanks,
Lieven
