www.digitalmars.com         C & C++   DMDScript  

digitalmars.D.learn - Exception stack traces on Phobos2? (Win or Lin)

reply "Nick Sabalausky" <a a.a> writes:
Is there a way to get stack traces for exceptions on Phobos2? For either 
Windows (preferably), or for Linux.

-------------------------------
Not sent from an iPhone.
Sep 04 2010
parent reply "Nick Sabalausky" <a a.a> writes:
"Nick Sabalausky" <a a.a> wrote in message 
news:i5ufqs$1v26$1 digitalmars.com...

 Is there a way to get stack traces for exceptions on Phobos2? For either 
 Windows (preferably), or for Linux.


Oops, sorry, I completely forgot I asked the exact same thing here little 
more than a month ago :)

But, someone there reported they're getting them in Linux as long as they 
don't use -release, but on Linux, even with both -debug and -g, I'm just 
getting meaningless addresses, no names or anything.
Sep 04 2010
next sibling parent Brad Roberts <braddr puremagic.com> writes:
On 9/4/2010 3:14 PM, Nick Sabalausky wrote:

 "Nick Sabalausky" <a a.a> wrote in message 
 news:i5ufqs$1v26$1 digitalmars.com...

 Is there a way to get stack traces for exceptions on Phobos2? For either 
 Windows (preferably), or for Linux.


 
 Oops, sorry, I completely forgot I asked the exact same thing here little 
 more than a month ago :)
 
 But, someone there reported they're getting them in Linux as long as they 
 don't use -release, but on Linux, even with both -debug and -g, I'm just 
 getting meaningless addresses, no names or anything. 
 


Progress has been made, but it's slow in coming.  Sean just landed a rewrite of
the buggy and inefficient std.demangle as core.demangle, an important stepping
stone to adding that to the stack trace generator.  I'm investigating a couple
issues with it right now, in fact.

  See also: bug 1001

Later,
Brad
Sep 04 2010
prev sibling parent reply Jonathan M Davis <jmdavisprog gmail.com> writes:
On Saturday 04 September 2010 15:14:30 Nick Sabalausky wrote:

 "Nick Sabalausky" <a a.a> wrote in message
 news:i5ufqs$1v26$1 digitalmars.com...
 

 Is there a way to get stack traces for exceptions on Phobos2? For either
 Windows (preferably), or for Linux.


 
 Oops, sorry, I completely forgot I asked the exact same thing here little
 more than a month ago :)
 
 But, someone there reported they're getting them in Linux as long as they
 don't use -release, but on Linux, even with both -debug and -g, I'm just
 getting meaningless addresses, no names or anything.


Add

-L--export-dynamic

to your dmd build command, and you should get mangled names. Demangled names
are 
still in the works unfortunately, but it's better than nothing.

Personally, my real problem is the difficulty in figuring out which unittest an 
exception was thrown from when it's not an assert in the unit test itself which 
fails. But fixing that would require name unit tests, and Walter hasn't agreed
to 
do those yet...

In any case, you should be able to get mangled names on linux.

- Jonathan M Davis
Sep 04 2010
parent reply "Nick Sabalausky" <a a.a> writes:
"Jonathan M Davis" <jmdavisprog gmail.com> wrote in message 
news:mailman.95.1283646171.858.digitalmars-d-learn puremagic.com...

 On Saturday 04 September 2010 15:14:30 Nick Sabalausky wrote:

 "Nick Sabalausky" <a a.a> wrote in message
 news:i5ufqs$1v26$1 digitalmars.com...


 Is there a way to get stack traces for exceptions on Phobos2? For 
 either
 Windows (preferably), or for Linux.


 Oops, sorry, I completely forgot I asked the exact same thing here little
 more than a month ago :)

 But, someone there reported they're getting them in Linux as long as they
 don't use -release, but on Linux, even with both -debug and -g, I'm just
 getting meaningless addresses, no names or anything.


 Add

 -L--export-dynamic

 to your dmd build command, and you should get mangled names. Demangled 
 names are
 still in the works unfortunately, but it's better than nothing.

 Personally, my real problem is the difficulty in figuring out which 
 unittest an
 exception was thrown from when it's not an assert in the unit test itself 
 which
 fails. But fixing that would require name unit tests, and Walter hasn't 
 agreed to
 do those yet...

 In any case, you should be able to get mangled names on linux.


Didn't work for me on Linux/2.048.
Sep 04 2010
parent reply Jonathan M Davis <jmdavisprog gmail.com> writes:
On Saturday 04 September 2010 17:25:23 Nick Sabalausky wrote:

 Didn't work for me on Linux/2.048.


I'm not sure what to tell you. I'm using dmd 2.048 on linux, and if I compile 
the following program

void func()
{
    assert(1 == 0);
}

void main()
{
    func();
}


with the command

dmd d.d -L--export-dynamic


I get

core.exception.AssertError d(3): Assertion failure
----------------
./d(_d_assertm+0x16) [0x805ae46]
./d(_D1d8__assertFiZv+0x12) [0x8058912]
./d(_D1d4funcFZv+0xd) [0x80588f1]
./d(_Dmain+0x8) [0x80588fc]
./d(_D2rt6dmain24mainUiPPaZi7runMainMFZv+0x1a) [0x805b036]
./d(_D2rt6dmain24mainUiPPaZi7tryExecMFMDFZvZv+0x24) [0x805af90]
./d(_D2rt6dmain24mainUiPPaZi6runAllMFZv+0x32) [0x805b07a]
./d(_D2rt6dmain24mainUiPPaZi7tryExecMFMDFZvZv+0x24) [0x805af90]
./d(main+0x96) [0x805af36]
/opt/lib32/lib/libc.so.6(__libc_start_main+0xe6) [0xf7557c76]
./d() [0x8058831]


So, it _should_ work on liunx with dmd 2.048.

- Jonathan M Davis
Sep 04 2010
parent reply "Nick Sabalausky" <a a.a> writes:
"Jonathan M Davis" <jmdavisprog gmail.com> wrote in message 
news:mailman.96.1283648486.858.digitalmars-d-learn puremagic.com...

 On Saturday 04 September 2010 17:25:23 Nick Sabalausky wrote:

 Didn't work for me on Linux/2.048.


 I'm not sure what to tell you. I'm using dmd 2.048 on linux, and if I 
 compile
 the following program


...

 So, it _should_ work on liunx with dmd 2.048.


It is working now, thanks. The problem turned out to be rdmd refusing to 
actually rebuild when using the -of switch and adding -L--export-dynamic.
Sep 04 2010
parent "Nick Sabalausky" <a a.a> writes:
"Nick Sabalausky" <a a.a> wrote in message 
news:i5v7gv$226$1 digitalmars.com...

 "Jonathan M Davis" <jmdavisprog gmail.com> wrote in message 
 news:mailman.96.1283648486.858.digitalmars-d-learn puremagic.com...

 On Saturday 04 September 2010 17:25:23 Nick Sabalausky wrote:

 Didn't work for me on Linux/2.048.


 I'm not sure what to tell you. I'm using dmd 2.048 on linux, and if I 
 compile
 the following program


 ...

 So, it _should_ work on liunx with dmd 2.048.


 It is working now, thanks. The problem turned out to be rdmd refusing to 
 actually rebuild when using the -of switch and adding -L--export-dynamic.


http://d.puremagic.com/issues/show_bug.cgi?id=4814
Sep 04 2010