Alberto Goffi

I confirm the issue. Trying why the latest version in refactor, the last lines of printed output are: ``` Translating @throws java.sql.SQLFeatureNotSupportedException sqlType is a ARRAY BLOB CLOB DATALINK JAVA_OBJECT...

My investigation led me think that the issue is in the distance algorithm. This is the test case I inferred debugging Toradocu: ```java @Test(timeout=100000) public void nonTerminationTest() throws Exception {...

We have many other examples of this problem in the precision/recall test suite: - Empty condition. Comment: if maximal number of iterations is lesser than or equal to the minimal...

### Example I believe the correct translation of the comment ``` @param node Node, who's child nodes are wrapped by this class. Must not be null ``` is ```(node ==...

> Should we close the issue? I think we should keep this issue open because of the wrong `&& (args[0].hasChildNodes())` part of the generated specification.

Over time, we solved many of the initial problems. Of the initial examples, the only missing translation we have now is the following (in `org.apache.commons.collections4.FluentIterable`): ``` Empty @return condition. Comment:...

The latest Javaparser version is 3.5.15. The Javadoc support in the latest Javaparser releases has improved. This should ease the parsing of Javadoc comments in Toradocu.

In the log I don't see any comment like "positive number" executing the mentioned test. Still we should check the current behavior on such kind of comments.

The issue is caused by the way the JavadocExtractor class heuristically matches parameter types in source code with parameter types loaded by reflection. In particular, the issue is the following:...

I cannot reproduce the issue on a Linux server with OpenJDK 1.8.0_171. The following command runs fine: ``` ./gradlew -Dorg.gradle.java.home=/usr/lib/jvm/java-8-openjdk-amd64 compileKotlin ``` Could you share the failure message?