working around the deprecation of public URL constructors (JDK-8294241)

[Michael McMahon]

Right, as you say using a relative path like that does not work for URL. 
It is possible to create relative URLs with the URL class (which is 
non-standard, unspecified behavior and one of the reasons why URI is 
better). So, I think we'd need to see the actual URL string being used.

- Michael.

On 16/12/2022 10:06, Daniel Fuchs wrote:
> Hi Rick,
>
> new URL("extinout/mybackup") throws a MalformedURLException too, so
> I wouldn't expect new URI("extinout/mybackup").toURL() to work.
>
> In that case I would recommend resolving the relative URI against
> its absolute base before translating to URL.
>
> My recommendation would be to use URL only at the point where
> you call openConnection().
>
>
> best regards,
>
> -- daniel
>
>
> On 15/12/2022 23:53, Rick Hillegas wrote:
>> Thanks for the quick response, Alan. One comment inline...
>>
>> On 12/15/22 12:24 PM, Alan Bateman wrote:
>>> On 15/12/2022 19:14, Rick Hillegas wrote:
>>>>
>>>> I see many warnings when I build Apache Derby's code and javadoc 
>>>> with Open JDK build 20-ea+27-2213. In particular, I see a number of 
>>>> warnings introduced by the following change:
>>>>
>>>> - JDK-8294241: Deprecate URL public constructors
>>>>
>>>> I naively tried to fix these warnings by changing instances of
>>>>
>>>>    new URL(urlString)
>>>>
>>>>
>>>> to
>>>>
>>>>    (new URI(urlString)).toURL() >>>
>>>>
>>>> Unfortunately, this breaks the Derby code. When I test Derby, I see 
>>>> errors like the following:
>>>>
>>>> java.lang.IllegalArgumentException: URI is not absolute
>>>>     at java.base/java.net.URL.of(URL.java:854)
>>>>     at java.base/java.net.URI.toURL(URI.java:1144)
>>>>     at 
>>>> org.apache.derby.impl.store.raw.RawStore.backup(RawStore.java:641)
>>>>
>>>> Can you give me advice about the best way to workaround the 
>>>> deprecation of the public URL constructors? I am looking for a 
>>>> solution which handles relative URL strings.
>>>>
>>> URL::toURL is specified to throw if the URI is not absolute although 
>>> the IAE may be a bit surprising as there isn't any arguments.
>>>
>>> Can you give an example of the "relative URL strings" so we can at 
>>> least see which URI scheme this is.
>>
>> This happens when urlString is a directory spec relative to the 
>> working directory of the JVM. E.g.,
>>
>>     extinout/mybackup
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.openjdk.org/pipermail/net-dev/attachments/20221228/7e3d6e99/attachment-0001.htm>