RFR: JDK-8290126: Add a check in JavadocTester for "javadoc should not crash"

[Jonathan Gibbons]

On Wed, 20 Jul 2022 19:52:34 GMT, Pavel Rappo <prappo at openjdk.org> wrote:

>> test/langtools/jdk/javadoc/lib/javadoc/tester/JavadocTester.java line 494:
>> 
>>> 492:         if (Files.exists(outputDir)) {
>>> 493:             if (automaticCheckNoStacktrace) {
>>> 494:                 checkOutput(Output.STDERR, false, " at com.sun.");
>> 
>> What is the reason for assuming package `com.sun` would be present in the stack trace? I notice that it is not present in the stack trace reported for the bug [JDK-8287379](https://bugs.openjdk.org/browse/JDK-8287379) which was recently fixed. Maybe we could additionally check for absence of boilerplate code like "An internal exception has occurred" or "Please file a bug against the javadoc tool ..."?
>
>> What is the reason for assuming package `com.sun` would be present in the stack trace? I notice that it is not present in the stack trace reported for the bug [JDK-8287379](https://bugs.openjdk.org/browse/JDK-8287379) which was recently fixed.
> 
> That's one reason why I suggested test-driving the new check by retrofitting it in the four existing use cases. Note that doclet/testSnippetTag/SnippetTester.java:55 looks for matches of:
> 
>     Pattern.compile("\\s*at.*\(.*\.java:\\d+\)")
> 
> While that pattern matches those exceptions from JDK-8287379, it might be too broad for the new check.

Ooops.  I was (incorrectly) thinking of the old days, `com.sun.tools.javadoc` and `com.sun.tools.doclets`.

-------------

PR: https://git.openjdk.org/jdk/pull/9458