On 14/04/15 20:33, Alan Bateman wrote:
....
This looks okay to me although maybe we can use the opportunity to
replace the use of StringTokenizer with a regex.
Right. I used String.split("\\|"), but reverted to the original code as
the regex doesn't use the fast-path. But you are right split is probably
better.
One comment on the javadoc is that it says "the constructors finds the
value of the system property", I assume this should be "reads" rather
than "finds".
Thanks. This was a typo in the original spec. Fixed.
> Also you can use {@code ...} in the blockquote to avoid
the escaping.
Done. We have lost the italic font, but that should be fine.
The above changes are capture here
http://cr.openjdk.java.net/~chegar/8075139/webrev.01/
Paul,
>
> Yes, one could use a Pattern.splitAsStream, sorry could not resist
:-) assuming this area is not sensitive to bootstrap issues e.g.
>
> hander =
p.splitAsStream().map(String::trim).flatMap(this::getHandler).findFirst().orElse(null);
With the restriction we now have on overriding "core" protocol handlers
we can actually use lambda's in this specific code path. I built a
webrev based on your suggestion. Not too bad. I marginally favor this
over webrev.01 above.
http://cr.openjdk.java.net/~chegar/8075139/webrev.02/src/java.base/share/classes/java/net/URL.java.sdiff.html
-Chris.
