debugger unusable with pallada and JNI

Hello, I was trying to convince a new company to use IDEA,
but they had terrible problems with the debugger in #2002
Is ist worth trying 4.0.3 or 2008 instead?

The following problems:
- IDEA stops at old breakpoints, which are not shown
in "view breakpoints" any more, and does not stop those breakpoints that are shown
- values from watches shows wrong (outdated) values !!

They were debugging JNI where java objects were created
from the C side, and when a certain class was instantiated
for the second time from C, all the values were still
shown from the first one in the debugger watches.

Maybe these problems only occur when JNI is involved?

7 comments
Comment actions Permalink

Pallada is very early stages of development. If you're trying to show IDEA to anyone, please use the released (4.0.3) version.

For a while, Pallada wil be just for the brave and strong of heart :)

0
Comment actions Permalink

Bug with breakpoints fixed in 2012.
Watches are updated every time application stops at breakpoint or step event. Could you please describe in detail how did you get the wrong values in watches?

Where was that value stored : local variable/parameter/field... What were you doing : stepping through code or you created new instances by evaluating some expression?

0
Comment actions Permalink

values are stored in fields in MapHeader:
see code below. the field min of MapHeader shows
the wrong (old) value when i>0 (still shows value of first created instance)
See the commented code that creates a dummy new instance from java -> then it works correctly!

for (byte i = 0; i < 8; i++)
{
//hdr = new MapHeader(); // then shows correct values when debugging!
hdr = dbal.getMapHeader(i);
long min = hdr.getMin();
log.debug("#"i"min="min"parcelwidth="+header.getParcelWidth());
}
-> getMapHeader is native method

0
Comment actions Permalink

stepping through the code, having breakpoint inside
the for loop, and hdr added to watches,
showing wrong values in watches-tab

0
Comment actions Permalink

one more problem with debbugger and hot swap:
on several windows xp machines I have experienced the
problem that when choosing "yes" to hot swap question
after compiling some classes while debugging,
IDEA hangs completely at the "Scanning Files" Dialog.
I have to kill the IDEA process then.
It does not happen always, but it effectively keeps
me from using that feature...

0
Comment actions Permalink

Could you, please, send the stacktrace next time you experience this problem?

It seems that JDI sometimes do not respond after redefineClasses method invocation, I'll make that IDEA will not hang in such cases, however no guarantee that debugee will continue working properly..

0
Comment actions Permalink

I do it all the time works like a charm in 4.0.3

Regards
Alex Oscherov
"Michael Damberger" <michael.damberger@t-online.de> wrote in message
news:27567411.1082033500961.JavaMail.javamailuser@localhost...

Hello, I was trying to convince a new company to use IDEA,
but they had terrible problems with the debugger in #2002
Is ist worth trying 4.0.3 or 2008 instead?

>

The following problems:
- IDEA stops at old breakpoints, which are not shown
in "view breakpoints" any more, and does not stop those breakpoints that

are shown

- values from watches shows wrong (outdated) values !!

>

They were debugging JNI where java objects were created
from the C side, and when a certain class was instantiated
for the second time from C, all the values were still
shown from the first one in the debugger watches.

>

Maybe these problems only occur when JNI is involved?

>
>


0

Please sign in to leave a comment.