I believe this to be a bug / edge case in the Uri.plusName(...) method...

Return a new Uri with a single path name appended to this Uri.

If I want to add foo?bar as a URI path segment I can do this:

uri := `/root/`.plusName("foo?bar")
echo(uri)         // --> /root/foo?bar    - would expect "/root/foo\?bar"
echo(uri.encode)  // --> /root/foo%3Fbar

The .toStr() method doesn't give quite what I'd expect, but at least the .encode() method does.

But should I try to encode foo/bar I get:

uri := `/root/`.plusName("foo/bar")
echo(uri)         // --> /root/foo/bar    - would expect "/root/foo\/bar"
echo(uri.encode)  // --> /root/foo/bar    - would expect "/root/foo%2Fbar"

For all intents and purposes of adding a single path name, I don't think .plusName() quite works as intended.

---

Looking for a work-around, I'm finding some inconsistent behaviour and I'm not sure which is correct...

uri1 := `/root/foo:bar/wotever`
uri2 := `/root/`.plusName(Uri.escapeToken("foo:bar", Uri.sectionPath), true).plusName("wotever")

echo(uri1.path)  // --> [root, foo:bar, wotever]
echo(uri2.path)  // --> [root, foo\:bar, wotever]

As you can see, one of the paths is escaped and the other isn't.

I've a feeling the difference could be purely academic - but I figure it's worth noting.

Ticket promoted to #2803 and assigned to brian

That is a little tricky, probably what you are suggesting is the correct behavior. Need to really dig in to think about it though.

If we did anything then we would call sys::Uri.escapeToken using sectionPath on the string. But changing it now might break code that might have been including special chars (even if that was incorrect).