I realised there’s a simpler solution to the problem. Instead of using a Set to 
track packages with non-modular documentation we can just return the module 
name that matches our internal model in #checkLinkCompatibility.

I also changed the JBS summary to „javadoc fails to link to docs with 
non-matching modularity“ so it covers both cases.

New Webrev: http://cr.openjdk.java.net/~hannesw/8240169/webrev.01/ 
<http://cr.openjdk.java.net/~hannesw/8240169/webrev.01/>

Hannes


> Am 02.04.2020 um 16:06 schrieb Hannes Wallnoefer 
> <hannes.wallnoe...@oracle.com>:
> 
> Please review:
> 
> JBS: https://bugs.openjdk.java.net/browse/JDK-8240169
> Webrev: http://cr.openjdk.java.net/~hannesw/8240169/webrev.00/
> 
> This patch allows using external documentation even if it doesn’t match the 
> external library in terms of modularity, i.e. non-modular documentation can 
> be used for an modular library and vice versa. Instead of showing an error a 
> warning is issued. There is still a warning if code and documentation do not 
> match, but we use the check to tweak reference lookup so that we are still 
> able to link to the appropriate documentation.
> 
> I think that all relevant cases for combinations of modular and non-modular 
> code are covered by existing tests (some of which change with this patch 
> obviouly).
> 
> TestLinkOptionWithAutomaticModule.java covers using a jar file as automatic 
> or unnamed module, both of which cases were already supported as of 
> JDK-8212233, so no changes there.
> 
> TestLinkOptionWithModule.java covers all combinations of modular and 
> non-modular code and documentation. For the two tests that cover non-matching 
> combinations, I changed the expected return code to OK and added expected 
> output with the link HTML. The message about the mismatching documentation is 
> still there, but it is a warning instead of an error. 
> 
> I also added the JBS id to the @bug tag in TestLinkOptionWithModule.java as 
> well as in TestClassCrossReferences.java which is another test that contains 
> significant changes.
> 
> Thanks,
> Hannes

Reply via email to