Handle URISyntaxException

[som-snytt]

Minor tweak to minor code that constructs a URI/URL.

Follow-up to #10335 noticed while porting forwardly.

[SethTisue]

I don't understand the motivation here — did something change in JDK 20?

[som-snytt]

It's trying to handle failure of URL construction, but doesn't catch URISyntaxException which is thrown by new URI. I didn't check what sorts of failures it might see IRL from jar manifests (which is calling the method).

The comment on the method that uses it:

  /** Expand manifest jar classpath entries: these are either urls, or paths
   *  relative to the location of the jar.
   */

I'm too lazy ATM to revisit the jar spec and also find out who is expanding the manifest class path, but catching the declared exception plugs the rabbit hole. If constructing the URL fails, it tries the relative path.

[SethTisue]

Okay. And in what user-facing context might this code be reached?

[som-snytt]

OK, still within scope of laziness, I see this is ancient scala foo.jar code via paulp. I don't know whether anyone does that any more.

I had assumed incorrectly it was part of the rjolly scripting stuff for -usemanifestcp.

I know I was just looking at that jsr-223 for some reason, because I finally understood what jarlister was for.

Now I don't know what scala -usemanifestcp foo.jar does.

I also recently had a chance to laugh at test/script-tests/README.

(Also remembered that the issue with -javabootclasspath had me looking at the code. Oh, I laughed at the readme here.)

[SethTisue]

okay, scala foo.jar seems like sufficiently low stakes functionality that I don't need to get all paranoid. anyway, it's hard to see how this could be a bad change.