bug on using breakpoints+debugger

[jbb]

Hi,

Did you really mean "step over instr."? This steps over one line of machine 
code not one line of source code. For that you need the "Step over" 
instruction.

Maybe this is the problem, although I cannot reproduce this here. Either with 
"step over instr." or "Step over".

jbb

PS It has revealed a bug in kdevelop2.0's disassemble tab though :-)


On Mon, 14 May 2001 12:00, you wrote:
> Hi,
>
> In my 3 last mails I have asked someone to explain the problem I have
> with the conversion from a string to a double.
> Finally, the problem comes from the breakpoints. Here is the tests I
> have made to make this conclusion :
>
> I have repeated 5 times the line `num=atof("1");' in my code, then I
> have put a breakpoint at one line, and run the program. Wherever I put
> the breakpoint, when the debugger stop at the breakpoint, the lines
> before the breakpoint have put the good value in num. But if I do just
> one "step over instr.", the line of the breakpoint put "-NaN" in num.
> And if I redo one step, the next line put the good value in num.
>
> Then I tried the same thing by changing each line by
> `sscanf("1","%lf",&num);'. I have the same result, except the fact that,
> when "-NaN" is put in num by the line of the breakpoint, the next lines
> doesn't return the good value.
>
> Then I have tried the same things by puting a breakpoint at the line 2
> and 3 but I have use the "run all" insteed of "step over instr.". I have
> exactly the same results.
>
> My conclusion is that when teh debugger stop at a breakpoint, it
> disturbs the well working of the conversion from a string to a double at
> least on the line of the breakpoint.
> If it disturbs other processes it is a great bug of the
> debugger+breakpoint procedure !
>
> Happy to help the developpers of KDevelop !
>
> Eric
>
> PS : I have forgotten to indicate my configuration. I use the packages
> given by the Mandrake 8.0 : KDE 2.1.1, KDevelop 1.4.1, ...

-
to unsubscribe from this list send an email to kdevelop-request at kdevelop.org with the following body:
unsubscribe »your-email-address«