mockpp-devel — Developers mailing list

Re: [Mockpp-devel] OutBound vs. verify() bug

[Ewald A. <mo...@ew...>]

Hello,

> it seems I hit a bug in mockpp. :-(

seems you're right :-) but I am not sure how to handle the issue. The original 
cause lies rather within OutBound and not within ConstraintList. Changing the 
behaviour of ContraintList as you proposed might introduce other problems in 
existing sources  so I am thinking about implementing an alternative way to 
handle verify() in the end.

greets

-- 
Ewald Arnold, Germany
mailto:ewald at ewald-arnold dot de

[Mockpp-devel] OutBound vs. verify() bug

[BUCHMULLER N. <no...@ni...>]

Hi,

it seems I hit a bug in mockpp. :-(

The problem is that if I use an OutBound constraint in a
VisitableMockObject, then I cannot use verify() on the object. The
attached test program (vm.cpp) will fail, complaining that the return list
of the OutBound object is empty:

mendel@vger:~$ ./vm 
terminate called after throwing an instance of
'mockpp::AssertionFailedError'
what():  foo/parameter1 added constraint[0] does not match: <returns
[]> != 0
Aborted (core dumped)

I tried to investigate the problem, and I found that
ConstraintList::checkImmediateValue() is called _two_ times on each
constraint. The first call is performed in addActual() - unless constraint
checking is deferred until verify() call - and the second call in
verify(). It seems to me as erroneous, since the first call pops the
outbound value from the list, so the second call will surely fail.
(This does not matter for the 'normal' constraints - they can be
eval()'d many times -, that's why it went unnoticed till now.)

I created a patch (attached as mockpp-1.16.4-OutBound_verify_fix.diff)
that worked for me, but unfortunately I was not able to compile the unit
tests (and so I was unable to check if it does not break something else).

Can you please check if my reasoning is right, and check if the patch is
correct?

Best regards,

norbi

Re: [Mockpp-devel] Cross compile

[Ewald A. <ew...@ew...>]

Hello,

> Undefined symbols:
>   "vtable for std::basic_ostringstream<char, std::char_traits<char>,
> std::allocator<char> >", referenced from:
>       __ZTVSt19basic_ostringstreamIcSt11char_traitsIcESaIcEE
> $non_lazy_ptr in main.o
>   "operator delete(void*)", referenced from:
>       yylex()    in scanner.o
>       yylex()    in scanner.o
>       yylex()    in scanner.o

I tried to reproduce the problem but everything works for me. I still assume, 
there is some unintended difference in the two invocations. Could you check 
in the lines before the error message if there is libstdc++ included, similar 
to this:

 -ldl -lstdc++

And please send me more of the output before the error message.

greets

-- 
Ewald Arnold, Germany
mailto:ewald at ewald-arnold dot de
icq:227435282

[Mockpp-devel] Cross compile

[Daniel E. <eg...@ma...>]

I need to cross compile mockpp, but I'm having trouble.

I'm trying to compile with

./configure --host=i686-apple-darwin9
make clean
make

but I get

Undefined symbols:
  "vtable for std::basic_ostringstream<char, std::char_traits<char>,  
std::allocator<char> >", referenced from:
      __ZTVSt19basic_ostringstreamIcSt11char_traitsIcESaIcEE 
$non_lazy_ptr in main.o
  "operator delete(void*)", referenced from:
      yylex()    in scanner.o
      yylex()    in scanner.o
      yylex()    in scanner.o

Things work fine, when I use

./configure
make clean
make

This is strange, because the platform I'm on is

% uname -mprs
Darwin 9.0.0b5 i386 i386

Is there anything more I should pass to configure?


/Daniel

[Mockpp-devel] mockpp 1.13 available

[Ewald A. <ew...@ew...>]

 - new ReturnObjectListStub is handy when you want to
   return a list of objects and onConsecutiveCalls is
   not appropriate
 - OutBound and TriggeredOutbound take iterators to
   add input data

=2D-=20
Ewald Arnold, Germany
mailto:ewald at ewald-arnold dot de
icq:227435282

[Mockpp-devel] mockpp 1.12.0 available

[Ewald A. <mo...@ew...>]

 - new constraint TriggeredOutbound together with TriggerStub
   and ReturnAndTriggerStub provide the use of outbound values
   based on another incoming method parameter
 - some fixes to ministl due to warnings with g++ 4.x
 - cleanup documentation (api + handbook)

=2D-=20
Ewald Arnold, Germany
mailto:ewald at ewald-arnold dot de
icq:227435282

Re: [Mockpp-devel] Question for mockpp

[Ewald A. <mo...@ew...>]

Hello,

> // the fucntion return user obj whose indendifier = id
>  int  find(  int id, User & usr);
> }
>
> i have no way to mock 'find' method with Chainable Mock Object, like,
>
> find_chainer.expect(once).with(eq......).

if I got it right you need to return a value by reference based on another 
parameter to the same method:

  id=1 ==> usr="one"
  id=2 ==> usr="two"

Currently this is not possible (besides of using a handcrafted method and some 
ReturnObjectList's :-)

There is a class Outbound which could return the value. But it returns one 
after the other, similar to ReturnObjectList. And the decision which 
expectation matches and which object to return is based on all parameters. So 
a potential solution might be to devise a class TriggerOutbound which somehow 
fills in the data when the method is actually invoked (after the parameters 
have been evaluated). Unfortunately all of the objects are completely 
independent of each others. So there must be some kind of bridge to the 
actual variable.

So the solution could look like this:

   Reference<User> *reference= new Reference<User>();
   find_chainer.expect(once())
                      .with(eq(1),
                               reference) 
                      .will(returnValue(true))
                      .will(triggerOutBound(reference))

I will think about this a bit more. Maybe I can find  a nice solution :-)

greets

-- 
Ewald Arnold, Germany
mailto:ewald at ewald-arnold dot de
icq:227435282

Re: [Mockpp-devel] Virtual function

[Ewald A. <ew...@ew...>]

Hello,

unfortunately I have no perfect solution. You always need some way to switch 
between production and test. That's probably what you have to pay for 
testability in C++. From my experience you have quite a lot of "unneeded" 
virtuals in the end. Everything else tends to become a bit tricky and more 
verbose.

What is you actual problem? Performance is usually none and security in the 
meaning of preventing to override the method is not safe.

Another solution I can offer is to turn B into a template. Instantiate one for 
production using A and one for testing using mock_A. 

There might as well be some other way of redirection. For example create a 
wrapper template for A/mock_A which helps you to select the correct one:

template <class A_type>
class A_wrapper
{
   A_type *a;
 
  A_wrapper (A_type *an_a)
    :  a(an_a)
    {}

   void test()
   {
     a->A_type::test();
   }
};

Maybe a compiler switch is acceptable which includes a "virtual" while 
compiling debug code:

#ifdef DEBUG
#define DEBUG_VIRTUAL virtual
#else
#define DEBUG_VIRTUAL 
#endif

hope that helps a bit :-)

> Hi,
>
> I have the following code :
>
> class A {
>     public:
>       void test()
>      {
>             // do some stuff
>       }
> }
>
> class A_mock :  public A,  public MOCKPP_NS::MockObject
> {
>    public:
>         A_mock() : ex("counter", 0){};
>
>         void test(
>               ec.inc();
>         );
>
>        ExpectationCounter ex;
> }
>
> class B
> {
>    public :
>         A*     m_A;
>         void f1()
>        {
>                 m_A->test();
>         }
> }
>
> int main ()
> {
>           A_mock*   mock = new A_mock();
>           mock->ex.setExpected(1);
>
>
>           B varB;
>           varB.m_A = mock;
>
>           mock->verify();
> }
>
>
>
> Since the test function in the base class A is NOT virtual, polymorphism
> don't work.  To work, I have to declarw A::test virtual.       My problem,
> is that in my design that function (A::test) should not be a virtual one.
> I find it akward to declare it virtual only for testing.   Any hint on
> doing that kind of test with mockpp?  I don't want to modify the B class.
>
> Thank you and great work on mockpp
>
> Patrick

-- 
Ewald Arnold, Germany
mailto:ewald at ewald-arnold dot de
icq:227435282

[Mockpp-devel] Virtual function

[Patrick F. <pat...@gm...>]

Hi,

I have the following code :

class A {
    public:
      void test()
     {
            // do some stuff
      }
}

class A_mock :  public A,  public MOCKPP_NS::MockObject
{
   public:
        A_mock() : ex("counter", 0){};

        void test(
              ec.inc();
        );

       ExpectationCounter ex;
}

class B
{
   public :
        A*     m_A;
        void f1()
       {
                m_A->test();
        }
}

int main ()
{
          A_mock*   mock = new A_mock();
          mock->ex.setExpected(1);


          B varB;
          varB.m_A = mock;

          mock->verify();
}



Since the test function in the base class A is NOT virtual, polymorphism
don't work.  To work, I have to declarw A::test virtual.       My problem,
is that in my design that function (A::test) should not be a virtual one.
I find it akward to declare it virtual only for testing.   Any hint on doing
that kind of test with mockpp?  I don't want to modify the B class.

Thank you and great work on mockpp

Patrick

[Mockpp-devel] mockpp-1.11.3: fixes for 64bit problems

[Ewald A. <mo...@ew...>]

-- 
Ewald Arnold, Germany
mailto:ewald at ewald-arnold dot de
icq:227435282

[Mockpp-devel] Patch and files to create mockpp-config

[Carle <Car...@US...>]

Hi,

Here's a small contribution to Mockpp that creates a "mockpp-config" scri=
pt to
help dealing with configure detection of Mockpp when used in other projec=
t
(install path & required includes and
libs). Since I'm not an expert in those kind of scripts, feel free to mod=
ify
anything to fit Mockpp more specific needs.

Patch file have been created with latest cvs development tree (Feb 26 200=
6). The
two files must be copied in the root directory. Patch can be applied by d=
oing
the following command in root directory :

patch -p1 < diffMockpp.patch

Afterward, configure needs to be regenerated.

If possible, I would appreciate if this patch could be merged to the curr=
ent cvs
development tree.

Thanks for Mockpp, it is very useful :-)

Carle

--=20
Carle Cote (http://pages.infinit.net/cotc2204/)

Laborius - Universit=E9 de Sherbrooke (http://www.gel.usherb.ca/laborius/=
)
MARIE's project homepage (http://marie.sf.net)

[Mockpp-devel] Re: Compilation with MINGW

[Ewald A. <ew...@ew...>]

Hello Jean-Pierre,

>d:/dve/tools/mockpp/mockpp-1.11.1/mockpp/examples/mock_greeter.cpp:339:
>undefined reference to `CxxTest::Link::active() const'
>
>
>I don't know how to find the method CxxTest::Link::active().
>IS it in CPPunit ? (installed on my machine)
>
>  
>
I just verified with cygwin. I had the strange problem that the wrapper 
executable did not compile due to missing prototypes for strlen and the 
like. But after tweaking libtool by adding "#include <string.h>" 
everything compiled and ran flawlessly. Mingw and cygwin share the same 
basis, so this is a bit puzzeling for me. I will try to install and 
verify with mingw environment.

In the meanwhile I suggest to run run-configure-ascii.sh to setup the 
files. This is what I always do. You could also add CXXTEST_EXPORT to 
the sources to enforce exporting the symbols. I am not experienced in 
mingw but windows platforms need exported symbols whereas Linux does 
not. I have no idea what happens in mixed environments. All I can say 
is, BCB5 and MSVC are working.

There is also a script bug-report.sh which could tell me more about your 
problems. Try running it.

greets

Re: [Mockpp-devel] a patch to mockpp-1.11.1 on sun studio 10

[Ewald A. <ew...@ew...>]

Hello,

> Recently I have to use mockpp-1.11.1 on sun studio 10.  In company with
> Emil, my colleague, we build it successfully after revising a few source.
> We have upload the patches to sourceforge and any comment is welcome.

thanks for the patches. The two patches for the counted 
stuff were useless, as these files are generated. And there 
are 12 places which needed changes. Maybe the patches were 
only intended to show the problem. But anyway, I changed 
the generator scripts. There is now a switch for sun to 
select the needed code. Most compilers insist on the 
additional "template". 
 
#ifdef __SUN__ 
#define MOCKPP_NO_TEMPLATE_HINT 
#endif 
 
Can I send you a tarball to verify the build on sun before 
I upload a new distribution? 

greets

-- 
Ewald Arnold, Germany
mailto:ewald at ewald-arnold dot de
icq:227435282

[Mockpp-devel] a patch to mockpp-1.11.1 on sun studio 10

[Linzhi h. <op...@ho...>]

Hi!

Recently I have to use mockpp-1.11.1 on sun studio 10.  In company with 
Emil, my colleague, we build it successfully after revising a few source. 
We have upload the patches to sourceforge and any comment is welcome.

Best Regards
Kylix Huang

_________________________________________________________________
免费下载 MSN Explorer:   http://explorer.msn.com/lccn  

[Mockpp-devel] Re: Help needed to compile with egcs-2.91.66

[Ewald A. <mo...@ew...>]

Hello Vincent,

> The compiler runs out of memory while trying to compile
> InvocationMocker.h.

Probably I had a similar problem with gcc 2.95. I don't remember which file it 
was but I stopped compilation after quite a while. There seems to be an 
optimization problem. When I used -O1 it worked fine. So I suggest to add -O1 
to your CXXFLAGS or add a single statement for the cpp file which breaks.

Is there a special reason why you use a rather old compiler? gcc 2.95 compiles 
the complete lib most of the time, but sometimes it crashes for no obvious 
reason. mockpp uses templates a lot so you should use a rather up-to-date 
compiler.

greets

-- 
Ewald Arnold, Germany
mailto:ewald at ewald-arnold dot de
icq:227435282

-- 
Ewald Arnold, Germany
mailto:ewald at ewald-arnold dot de
icq:227435282

-- 
Ewald Arnold, Germany
mailto:ewald at ewald-arnold dot de
icq:227435282

[Mockpp-devel] mockpp-1.8.0 available

[Ewald A. <ew...@ew...>]

A new section in the handbook about optimizing production code for testing=
=20
purposes has been started. There is also a sample implementation on how to=
=20
handle time dependencies.

2005-05-06  1.8.0:

 - included missing project file for bcb5
 - fix for msvc7.1
 - cleanup ThrowableList
 - basic support for "Design By Contract"
 - time server classes
 - extended handbook

=2D-=20
Ewald Arnold, Germany
mailto:ewald at ewald-arnold dot de
icq:227435282

[Mockpp-devel] Re: mockpp 1.4.0's compiling problem

[Ewald A. <mo...@ew...>]

Hello,

> I tried to compile mockpp 1.4.0 on Linux with gcc 2.9.5.

I am impressed that gcc 2.95 even works somehow ;-) I don't remember when I 
tried last but as far as I rember it crashed due to am internal error. The 
usage of the templates needs a really capable and up-to-date compiler ;-) 
None of the MS compilers I know of does it right.

> > TrackingCounter.cpp:165: `::find' undeclared (first use here)
> I think you forgot to include <algorithm> for find.

did you add the #include and did it work then? If you really got 1.2.x working 
there is nothing against it. Obviously there is some hidden compiler 
dependency. I will include the line and release 1.5.0 the next days with some 
enhancements for VisitableMockObjects.

greets

-- 
Ewald Arnold, Germany
mailto:ewald at ewald-arnold dot des
icq:227435282

[Mockpp-devel] mockpp-1.2.0 final available

[Ewald A. <mo...@ew...>]

Fixed all known bugs and issues in the sources. Documentation cleanup. See 
ChangeLog for details. 

-- 
Ewald Arnold, Germany
mailto:ewald at ewald-arnold dot de
icq:227435282

[Mockpp-devel] mockpp 1.2.0beta coming soon

[Ewald A. <mo...@ew...>]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hello out there,

Currently there is a port of jMock in the works. This will offer another
MockObject approach which removes several limitations with the current
implementation.

Download from cvs if you are interested.

 See http://jmock.org/getting-started.html for an introduction into the java
original.

- --
Ewald Arnold, Germany
mailto:ewald at ewald-arnold dot de
icq:227435282
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.2-rc1-SuSE (GNU/Linux)

iD8DBQFBQxlRC5JDB2jxrzQRAmlTAJ47KULjgxMqgz+o+b++jTW8ZxTuowCgrh2p
MgQDI7u8gPrio59AtBnADkA=
=uQQi
-----END PGP SIGNATURE-----

[Mockpp-devel] FASD project: Online survey launched

[<ben...@id...>]

Dear Open Source developer

I am doing a research project on "Fun and Software Development" in which I kindly invite you to participate.
You will find the online survey under http://fasd.ethz.ch/qsf/. The questionnaire consists of 53 questions and you will need about 15 minutes to complete it.

With the FASD project (Fun and Software Development) we want to define the motivational significance of fun when software developers decide to engage in Open Source projects. What is special about our research project is that a similar survey is planned with software developers in commercial firms. This procedure allows the immediate comparison between the involved individuals and the conditions of production of these two development models. Thus we hope to obtain substantial new insights to the phenomenon of Open Source Development.


With many thanks for your participation,
Benno Luthiger


PS:
The results of the survey will be published under http://www.isu.unizh.ch/fuehrung/blprojects/FASD/.
We have set up the mailing list fa...@we... for this study. Please see http://fasd.ethz.ch/qsf/mailinglist_en.html for registration to this mailing list.

_______________________________________________________________________

Benno Luthiger
Swiss Federal Institute of Technology Zurich
8092 Zurich

Mail: benno.luthiger(at)id.ethz.ch
_______________________________________________________________________

[Mockpp-devel] mockpp 1.1.8 released

[Ewald A. <mo...@ew...>]

=2D----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Mockpp 1.1.8 is released=20

=2D - Port to OpenBsd 3.1 and g++ 2.95
=2D - fix for MSVC
=2D - simplified MockObject verify()
=2D - removed useless and buggy multithread stuff
=2D - fixed MockObject::verify()

=2D From my personal knowledge mockpp is now running on
=2D - Linux (SuSE 8.2) + gcc 3.3
=2D - OpenBsd 3.1 + gcc 2.95.3
=2D - Windows (ME, NT) + Borland CBuilder 5.=20

Since OpenBSD is a working platform I assume that other *BSD
flavours work fine as well.

There is also a report about working MS Visual C++ though I have no
project files yet.

=2D --=20
Ewald Arnold, Germany
mailto:ewald at ewald-arnold dot de
mobil/sms:+49-162-8001240
=2D----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.2-rc1-SuSE (GNU/Linux)

iD8DBQE+maSmXyQcrD0ZXUYRAjfoAJ9tksP8Kp5/wVYBL4yjoHCCMbN92ACfWuK+
BOpf2GMJ1mUIJCH22+Q+oVc=3D
=3D3b+O
=2D----END PGP SIGNATURE-----

[Mockpp-devel] Few questions

[Philippe F. <ph...@fr...>]

	Hi,

Something along the lines of mockpp has been long expected. I was thinking 
about writing something simpler.

I have a few questions:

From reading the documentation, I understand that you can check on one 
method if it is called the right number of times with the right arguments. 
Is it possible to combine this for multiple methods ?

For example:
class A 
{
	virtual	void method1( int i );
	virtual	void method2( int j );
};

I want to check that my code calls
a.method1( 1 );
a.method2( 3 );

and I want an error if it does:
a.method2( 3 );
a.method1( 1 );

The arguments are correct, the number of call is correct but the methods 
have been called in reverse order.


Is a more integration with CppUnit planned ? It would be quite convenient to 
have more integration.

	regards,

	Philippe

[Mockpp-devel] mockpp 1.1.5

[Ewald A. <mo...@ew...>]

This bug fix release closes a memory leak and should now compile and run on=
=20
Windoze under BCB5.

=2D-=20
Ewald Arnold, Germany
mailto:ewald at ewald-arnold dot de
mobil/sms:+49-162-8001240

[Mockpp-devel] Re: [MO-java-dev] ANNOUNCE: mockpp final release

[Steve F. <ste...@m3...>]

Welcome to the club, but is there a URL?

Steve

Ewald Arnold wrote:
> The final release of mockpp is available.
> 
> mockpp is a platform independent generic unit testing framework for C++. It's 
> goal is to facilitate developing unit tests in the spirit of 
> Mock Objects (http://www.mockobjects.com) and EasyMock 
> (http://www.easymock.org).
> 
> Mockpp has no own testing framework. So CppUnit (http://cppunit.sf.net) is 
> recommended.
> 
> The main development platform of mockpp is Linux. But it is intended to be 
> platform independent and a port to Borland BCB is already in work. 
> 
> Contributions, ports to other platforms and comments are highly welcome.
> 
> greetings
> 

[Mockpp-devel] ANNOUNCE: mockpp final release

[Ewald A. <mo...@ew...>]

The final release of mockpp is available.

mockpp is a platform independent generic unit testing framework for C++. It=
's=20
goal is to facilitate developing unit tests in the spirit of=20
Mock Objects (http://www.mockobjects.com) and EasyMock=20
(http://www.easymock.org).

Mockpp has no own testing framework. So CppUnit (http://cppunit.sf.net) is=
=20
recommended.

The main development platform of mockpp is Linux. But it is intended to be=
=20
platform independent and a port to Borland BCB is already in work.=20

Contributions, ports to other platforms and comments are highly welcome.

greetings

=2D-=20
Ewald Arnold, Germany
mailto:ewald at ewald-arnold dot de
mobil/sms:+49-162-8001240