RE: [Japi] Issues in latest results

[Jeroen Frijters]

Stuart Ballard wrote:
> The (lack of) jdk10 results are my fault: I implemented @filename
> arguments to expect one argument per line, but when I changed my
> nightly script from `cat jdk10.pkgs` to @jdk10.pkgs I forgot to make
> sure the arguments actually *were* all on one line.
> 
> But I'm not so sure about one other aspect of the diffs. There seems
> to be something of an epidemic of reports of the staticness of inner
> classes being wrong. Jeroen, are you aware of any way your changes
> might have caused these to be calculated wrong?

Oops. I completely forgot about that. The class modifiers are stored in
two places (for inner classes) and only one of those is read eagerly.
This really sucks, because for getModifiers() to work (which is used by
Japize to determine if a class is interesting or not) I also need to
parse the rest of the file. That would make the entire exercise useless.

I'll add a isPublicOrProtected() method to ClassFile that Japize can
then use instead of getModifiers() and then fix getModifiers() to
trigger a parse of the rest of the file.

Regards,
Jeroen