Bugzilla – Full Text Bug Listing
| Summary: | [README] breakpoint set on "if(true)" line never hit | ||
|---|---|---|---|
| Product: | [Eclipse Project] JDT | Reporter: | Grant Gayed <grant_gayed> |
| Component: | Debug | Assignee: | JDT-Debug-Inbox <jdt-debug-inbox> |
| Status: | RESOLVED DUPLICATE | QA Contact: | |
| Severity: | normal | ||
| Priority: | P3 | CC: | markus.kell.r, Michael_Rennie, Olivier_Thomann |
| Version: | 3.7 | Keywords: | readme |
| Target Milestone: | --- | ||
| Hardware: | PC | ||
| OS: | Windows 7 | ||
| Whiteboard: | |||
This is working fine, albeit a bit confusing.
The reason the breakpoint is not hit is because the if statement is being in-lined by the compiler (thanks Olivier for confirming this) - for both javac and the Eclipse compiler.
For example the following snippet:
if (true) {
System.out.println("Hello World");
}
looks like the following in the class file:
0 getstatic java.lang.System.out : java.io.PrintStream [19]
3 ldc <String "Hello World"> [25]
5 invokevirtual java.io.PrintStream.println(java.lang.String) : void [27]
8 return
I thought this might be the case. However it's misleading to allow a user to set a breakpoint there, show it visually in the gutter, and then just ignore it when debugging. I found this because it's something I actually tried to do last week. Since it allowed me to set the bp there I guessed that the compiler optimization of the "if (true)" might somehow not apply when run in debug mode, and that the bp would be hit. |
I20110719-0800 - create a class with main method: public static void main(String[] args) { if (true) { int j = 0; } } - set a breakpoint on the "if (true) {" line - debug it, the breakpoint is never hit - set another breakpoint on the line below, debug it, note that this new breakpoint is hit, but the original one still is not