Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
> "Kevin Grittner" <Kevin(dot)Grittner(at)wicourts(dot)gov> writes:
>> Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
>>> We'd probably not want to apply this as-is, but should first
>>> tighten up what characters URLPath allows, per Kevin's spec
>>> research.
>
>> If we're headed that way, I figured I should double-check. The
>> RFC I referenced was later obsoleted by:
>> http://www.ietf.org/rfc/rfc3986.txt
>
> On reflection, since we're changing the behavior anyway, it seems
> like the most defensible thing to do is make the TS parser follow
> the RFC's allowed character set exactly. The newer RFC doesn't
> restrict '#' so that possible corner case is gone.
It seems worth mentioning that there is a BSD licensed URI parser on
sourceforge:
http://uriparser.sourceforge.net/
I'm not advocating for using it, I just ran across it and it seemed
of possible interest.
-Kevin