Details
Bug
Resolved
Major
Description
at least in class JavaMethod
public int hashCode() { int hashCode = name.hashCode(); if (returns != null) hashCode *= returns.hashCode(); hashCode *= getParameters().length; return hashCode; }
when parameter.set has zero length, hashcode turns to be zero as well
this affects usage in Maps
improvement would be to use:
http://commons.apache.org/lang/api-2.4/org/apache/commons/lang/builder/HashCodeBuilder.html
this could be also wrong in other classes
Activity
Lars K. W. Gohlke
added a comment - Lars K. W. Gohlke
added a comment -
public int hashCode() {
int hashCode = name.hashCode();
if (returns != null )
hashCode *= returns.hashCode();
hashCode *= getParameters().length;
return hashCode;
}
Robert Scholte
made changes -
| Field | Original Value | New Value |
|---|---|---|
| Description |
at least in class JavaMethod
{code] public int hashCode() { int hashCode = name.hashCode(); if (returns != null) hashCode *= returns.hashCode(); hashCode *= getParameters().length; return hashCode; } {code} when parameter.set has zero length, hashcode turns to be zero as well this affects usage in Maps improvement would be to use: http://commons.apache.org/lang/api-2.4/org/apache/commons/lang/builder/HashCodeBuilder.html this could be also wrong in other classes |
at least in class JavaMethod
{code} public int hashCode() { int hashCode = name.hashCode(); if (returns != null) hashCode *= returns.hashCode(); hashCode *= getParameters().length; return hashCode; } {code} when parameter.set has zero length, hashcode turns to be zero as well this affects usage in Maps improvement would be to use: http://commons.apache.org/lang/api-2.4/org/apache/commons/lang/builder/HashCodeBuilder.html this could be also wrong in other classes |
I'll review alle hashcodes, but I don't want to depend on commons-lang just for this.
Right now QDox has zero compile-time dependencies and I'd like to keep it that way.
But hashCodes which could result in 0 are indeed wrong.
Robert Scholte
added a comment - I'll review alle hashcodes, but I don't want to depend on commons-lang just for this.
Right now QDox has zero compile-time dependencies and I'd like to keep it that way.
But hashCodes which could result in 0 are indeed wrong. sum the hashcodes up, would be the first step and
in this special case the call signature with the full qualified classname would be good for hashing
Lars K. W. Gohlke
added a comment - sum the hashcodes up, would be the first step and
in this special case the call signature with the full qualified classname would be good for hashing Fixed in rev. 1293
Please verify it and close the issue if it is valid.
Robert Scholte
added a comment - Fixed in rev. 1293
Please verify it and close the issue if it is valid. Robert Scholte
made changes -
| Status | Open [ 1 ] | Resolved [ 5 ] |
| Assignee | Robert Scholte [ rfscholte ] | |
| Fix Version/s | 2.0 [ 15636 ] | |
| Resolution | Fixed [ 1 ] |
what about varargs in JavaParameter hashcode? Now (int ... x) and (int x) would result in the same hash
the rest is ok as I know.
Lars K. W. Gohlke
added a comment - what about varargs in JavaParameter hashcode? Now (int ... x) and (int x) would result in the same hash
https://fisheye.codehaus.org/browse/qdox/trunk/qdox/src/main/java/com/thoughtworks/qdox/model/DefaultJavaParameter.java?r=1293#to126
the rest is ok as I know. Fixed in rev. 1296
Just made me realize that I have to review all the equals() as well.
Robert Scholte
added a comment - Fixed in rev. 1296
Just made me realize that I have to review all the equals() as well. 