[foreign] RFR 8222919: jextract should compile generated java sources rather than use ASM to generate class files

Sundararajan Athijegannathan sundararajan.athijegannathan at oracle.com
Wed Apr 24 13:18:27 UTC 2019

Hi Jorn,

Yes, I see 3 test failures on Windows (only) with internal mach5 build. 
I'll check this out. Thanks.


On 24/04/19, 4:35 PM, Jorn Vernee wrote:
> Hi Sundar,
> In InMemoryJavaCompiler.FileManager::getJavaFileForOutput; Should this 
> use computeIfAbsent instead of put? You're probably more aware of the 
> backing implementation. Is there any chance a file with the same name 
> is requested twice, and then the previously created ClassFile object 
> being overwritten?
> Also, there are some tests failing. This seems to be due to Unix vs. 
> Windows path separators, for instance in the Runner test:
>   test Runner.testJarManifest(): failure
>   java.lang.AssertionError: Sets differ: expected [com.acme.pad_h, 
> com.acme.pad_h$anon$pad_h$1195, com.acme.pad_h$PaddyStruct] but got 
> [com\\acme\\pad_h$anon$pad_h$1195, com\\acme\\pad_h, 
> com\\acme\\pad_h$PaddyStruct]
>     at org.testng.Assert.fail(Assert.java:94)
> I'm looking into this right now, but maybe you know where the problem 
> might be?
> Cheers,
> Jorn
> Sundararajan Athijegannathan schreef op 2019-04-24 11:34:
>> Please review.
>> Bug: https://bugs.openjdk.java.net/browse/JDK-8222919
>> Webrev: https://cr.openjdk.java.net/~sundar/8222919/webrev.00/
>> Thanks
>> -Sundar

More information about the panama-dev mailing list