[not-in-charge]

[Affan Ahmed]

Hi,
I have looked up all the solutions for this problem, but nothing seems to
work for. I would really appreciate any help and a reason as to why this
happens...

When I complile I seem to get the following errors:

g++ -I. -g -Wall -fno-inline -c timers.cpp
g++ -I. -g -Wall -fno-inline -c tools.cpp
g++ -I. -g -Wall -fno-inline -c test-app.cpp
test-app.cpp: In member function `virtual int TestTimer3::Expire()':
test-app.cpp:63: warning: int format, handle arg (arg 3)
g++ timers.o tools.o test-app.o -o test-app
timers.o(.gnu.linkonce.d._ZTI8cTimeVal+0x8):/nfs/guest/asyed/ilenseCVS/timer
s/timers.cpp:102: undefined reference to `typeinfo for cSimTime'
timers.o(.gnu.linkonce.t._ZN8cSimTimeC2Ev+0x8): In function
`cSimTime::cSimTime[not-in-charge]()':
/nfs/guest/asyed/ilenseCVS/timers/timers.cpp:102: undefined reference to
`vtable for cSimTime'
timers.o(.gnu.linkonce.t._ZN8cSimTimeD2Ev+0xb): In function
`cSimTime::~cSimTime [not-in-charge]()':
/nfs/guest/asyed/ilenseCVS/timers/timers.cpp:262: undefined reference to
`vtable for cSimTime'
collect2: ld returned 1 exit status
make: *** [test-app] Error 1

I have the cSimTime class as a pure abstract class and use cTimeVal as
subclass implementing *all* the pure virtual functions. Here is the code:
(Note the first error comes when i am defining the first non-inline virutal
function of the timeVal class i.e. on line 102 of timers.cpp i have the
implementation of getLine())

class cSimTime{
public:
cSimTime() {};
virtual ~cSimTime() {};

//the current virtual time
virtual void getTime() = 0;

//return the current time in the cSimTime object
virtual cSimTime* returnCurrentTime()=0;

//return -1,0,1 for < == >
virtual short compareTime( cSimTime* compareWith)=0;

//add to this time object a specific # of virtual ticks
virtual void addTime(timeTicks ticks) =0;

//set time arbitrarily to the value passed
virtual void setTime(timeTicks ticks) =0;

//subtract from this time object y and return their value... return value
interpreted by implementor of subclass
virtual cSimTime* subTime(cSimTime* y) =0 ;
//generic printing funtion
virtual void print()=0;

};
/*One abstraction of the virtual time concept in the simulator
*
*For example refer to the test app.
*/
class cTimeVal: public cSimTime{

public:
cTimeVal(){};
~cTimeVal(){};

//the current virtual time
void getTime();

//return the current time in the cSimTime object
cSimTime* returnCurrentTime();

//return -1,0,1 for < == >
short compareTime(cSimTime* compareWith);

//add to this time object a specific # of virtual ticks
void addTime(timeTicks ticks);

//set time arbitrarily to the value passed
void setTime(timeTicks ticks) { tv.tv_sec = ticks;};

//subtract from this time object y and return their value... return value
interpreted by implementor of subclass
cSimTime* subTime(cSimTime* y);

//printing for the timeval struct
void print(){printf("time=%ld:%06ld \n",tv.tv_sec,tv.tv_usec);};

struct timeval tv;//the concept of time for this instance
};

////////////////////////////////////////////////////////////////////////////
////////////////////////////////////////////////////////////////

Please note that i am also creating inside subTime function a static
variable of type cTimeVal. I am not sure but i think this might be causing
some of the errors.

I would really appreciate if people could help me out of this.

Thanks.
Regards
Affan

[Malte Starostik]

Affan Ahmed schrieb:

Hi,
I have looked up all the solutions for this problem, but nothing seems to
work for. I would really appreciate any help and a reason as to why this
happens...

When I complile I seem to get the following errors:

timers.o(.gnu.linkonce.d._ZTI8cTimeVal+0x8):/nfs/guest/asyed/ilenseCVS/timer
s/timers.cpp:102: undefined reference to `typeinfo for cSimTime'
timers.o(.gnu.linkonce.t._ZN8cSimTimeC2Ev+0x8): In function
`cSimTime::cSimTime[not-in-charge]()':
/nfs/guest/asyed/ilenseCVS/timers/timers.cpp:102: undefined reference to
`vtable for cSimTime'
[snip]
I have the cSimTime class as a pure abstract class and use cTimeVal as
subclass implementing *all* the pure virtual functions. Here is the code:
(Note the first error comes when i am defining the first non-inline virutal
function of the timeVal class i.e. on line 102 of timers.cpp i have the
implementation of getLine())
This is a linker error that belongs in a a gcc-specific newsgroup.
Short answer: check if you really defined *all* virtual functions, this
kind of error is typical for something missing there.
After adding the neccessary #includes, a dummy typedef int timeTicks,
empty non-inline definitions for the other virtuals of cTimeVal and int
main() { cTimeVal t; } the code compiles and links fine for me.
class cSimTime{
public:
cSimTime() {};
virtual ~cSimTime() {}; ^ no semicolon after member function body

[snip]
struct timeval tv;//the concept of time for this instance

^^ struct keyword not needed in C++ in variable declarations

Wild guess: the paths in your make output suggest you're working on an
NFS mounted file system; make sure the system clocks on the NFS server
and client are in sync so make can get the dependencies right, but
that's really not topical here.

Regards,
Malte

[Affan Ahmed]

Thanks everybody.. I also found the prob.... note that my pure virtual
functions were not exactly pure... there was a space between () and =0!!!!!
gawd... this sucks.. i would have thought i would get some compiler errors
or warning s atleast!!!!
"Affan Ahmed" <as***@usc.edu> wrote in message
news:ck**********@gist.usc.edu...

Hi,
I have looked up all the solutions for this problem, but nothing seems to
work for. I would really appreciate any help and a reason as to why this
happens...

When I complile I seem to get the following errors:

g++ -I. -g -Wall -fno-inline -c timers.cpp
g++ -I. -g -Wall -fno-inline -c tools.cpp
g++ -I. -g -Wall -fno-inline -c test-app.cpp
test-app.cpp: In member function `virtual int TestTimer3::Expire()':
test-app.cpp:63: warning: int format, handle arg (arg 3)
g++ timers.o tools.o test-app.o -o test-app
timers.o(.gnu.linkonce.d._ZTI8cTimeVal+0x8):/nfs/guest/asyed/ilenseCVS/timer s/timers.cpp:102: undefined reference to `typeinfo for cSimTime'
timers.o(.gnu.linkonce.t._ZN8cSimTimeC2Ev+0x8): In function
`cSimTime::cSimTime[not-in-charge]()':
/nfs/guest/asyed/ilenseCVS/timers/timers.cpp:102: undefined reference to
`vtable for cSimTime'
timers.o(.gnu.linkonce.t._ZN8cSimTimeD2Ev+0xb): In function
`cSimTime::~cSimTime [not-in-charge]()':
/nfs/guest/asyed/ilenseCVS/timers/timers.cpp:262: undefined reference to
`vtable for cSimTime'
collect2: ld returned 1 exit status
make: *** [test-app] Error 1

I have the cSimTime class as a pure abstract class and use cTimeVal as
subclass implementing *all* the pure virtual functions. Here is the code:
(Note the first error comes when i am defining the first non-inline virutal function of the timeVal class i.e. on line 102 of timers.cpp i have the
implementation of getLine())

class cSimTime{
public:
cSimTime() {};
virtual ~cSimTime() {};

//the current virtual time
virtual void getTime() = 0;

//return the current time in the cSimTime object
virtual cSimTime* returnCurrentTime()=0;

//return -1,0,1 for < == >
virtual short compareTime( cSimTime* compareWith)=0;

//add to this time object a specific # of virtual ticks
virtual void addTime(timeTicks ticks) =0;

//set time arbitrarily to the value passed
virtual void setTime(timeTicks ticks) =0;

//subtract from this time object y and return their value... return value interpreted by implementor of subclass
virtual cSimTime* subTime(cSimTime* y) =0 ;
//generic printing funtion
virtual void print()=0;

};
/*One abstraction of the virtual time concept in the simulator
*
*For example refer to the test app.
*/
class cTimeVal: public cSimTime{

public:
cTimeVal(){};
~cTimeVal(){};

//the current virtual time
void getTime();

//return the current time in the cSimTime object
cSimTime* returnCurrentTime();

//return -1,0,1 for < == >
short compareTime(cSimTime* compareWith);

//add to this time object a specific # of virtual ticks
void addTime(timeTicks ticks);

//set time arbitrarily to the value passed
void setTime(timeTicks ticks) { tv.tv_sec = ticks;};

//subtract from this time object y and return their value... return value
interpreted by implementor of subclass
cSimTime* subTime(cSimTime* y);

//printing for the timeval struct
void print(){printf("time=%ld:%06ld \n",tv.tv_sec,tv.tv_usec);};

struct timeval tv;//the concept of time for this instance
};

//////////////////////////////////////////////////////////////////////////// ////////////////////////////////////////////////////////////////

Please note that i am also creating inside subTime function a static
variable of type cTimeVal. I am not sure but i think this might be causing
some of the errors.

I would really appreciate if people could help me out of this.

Thanks.
Regards
Affan

[Malte Starostik]

Affan Ahmed schrieb:

Thanks everybody.. I also found the prob.... note that my pure virtual
functions were not exactly pure... there was a space between () and =0!!!!!
gawd... this sucks.. i would have thought i would get some compiler errors
or warning s atleast!!!!

Out of curiosity, where exactly was that space?

struct A
{
virtual ~A() {}
virtual void f() = 0;
virtual void g()=0;
virtual void
h ( )
=
0
;
};

All of the above variations are syntactically correct declarations of
pure virtual functions.

Regards,
Malte

[Affan Ahmed]

well mine was
virtual void getTime() =0;
and when i removed the space the same code compiled perfectly. I had thought
the same .. i.e. it is syntactically correct, but browsing over the net for
similar problem showed that these errors were for a virtual function that
was *not* implemented... Since I had all defined all function in my
subclass, the only prob that finally seemed to remain was that my base class
pure virtual func was not being interpreted as pure virtual. So I checked it
again, and found this discrepency.
I am sure if some one can shed some light on this .. it would be a good
thing to know.

thanks.

Affan
"Malte Starostik" <ma*************@t-online.de> wrote in message
news:ck*************@news.t-online.com...

Affan Ahmed schrieb:

Thanks everybody.. I also found the prob.... note that my pure virtual
functions were not exactly pure... there was a space between () and =0!!!!! gawd... this sucks.. i would have thought i would get some compiler errors or warning s atleast!!!!

Out of curiosity, where exactly was that space?

struct A
{
virtual ~A() {}
virtual void f() = 0;
virtual void g()=0;
virtual void
h ( )
=
0
;
};

All of the above variations are syntactically correct declarations of
pure virtual functions.

Regards,
Malte

[Yevgen Muntyan]

Affan Ahmed wrote:

well mine was
virtual void getTime() =0;
and when i removed the space the same code compiled perfectly.

Are you sure? Could you take the file which compiles now, put _one_
space after the 'getTime()' and try to compile it again?
Or, alternatively, take the original file which does not compile
and remove _one_ space from it and compile?

If space really causes the problem, then please post contents
of the file and output of 'g++ --version'.

(Yet, there is something strange with spaces in your original post
(and this as well). Maybe it's a problem of my newsreader, or of
your mail client, but it's weird. Maybe this is a problem?)

Thank you,
Yevgen

[John Harrison]

"Affan Ahmed" <as***@usc.edu> wrote in message
news:ck**********@gist.usc.edu...

well mine was
virtual void getTime() =0;
and when i removed the space the same code compiled perfectly. I had
thought
the same .. i.e. it is syntactically correct, but browsing over the net
for
similar problem showed that these errors were for a virtual function that
was *not* implemented... Since I had all defined all function in my
subclass, the only prob that finally seemed to remain was that my base
class
pure virtual func was not being interpreted as pure virtual. So I checked
it
again, and found this discrepency.
I am sure if some one can shed some light on this .. it would be a good
thing to know.

Isn't this one of those problems that goes away when you force a
recompilation? Adding or removing spaces wasn't the fix. It was the fact
that by changing the header file you forced a recompilation of all the
source files that needed recompiling.

john

[Malte Starostik]

John Harrison schrieb:

"Affan Ahmed" <as***@usc.edu> wrote in message
news:ck**********@gist.usc.edu...

well mine was
virtual void getTime() =0;
and when i removed the space the same code compiled perfectly. I had
thought
the same .. i.e. it is syntactically correct, but browsing over the net
for
similar problem showed that these errors were for a virtual function that
was *not* implemented...

Isn't this one of those problems that goes away when you force a
recompilation? Adding or removing spaces wasn't the fix. It was the fact
that by changing the header file you forced a recompilation of all the
source files that needed recompiling.

I'd very much suspect that. If it really is the space then the only
thing I can imagine is the space being some Unicode non-breaking space
char, you could check that with a hex editor...

<totally OT>
If it was the forced recompilation that fixed it, then something is
wrong with your dependency info; check your Makefile and/or run make
with verbose output to see what files it considers in its dependency
checks. But as changing the header finally resulted in a recompilation,
my guess would really be a desynced file time issue as I mentioned
earlier. Never use a source tree on NFS without NTP.
</totally OT>

Regards,
Malte

[Affan Ahmed]

Well it is true that when i tried to recompile with a space in between ... I
didnt find any problem!... I am kinda uncomfortable with makefile and was
using another template which I tried to fill in as I thought was neccecary.
but i havent really changed the makefile as yet either. So the only
plausible thing could have been the NFS problem (which sadly i didnt really
graps). However if it was a time sync issue, then why did it ever get
corrected when i did remove the space?
Thanks any way.. as long as its not bugging me any longer.
Regards,
Affan

..

I'd very much suspect that. If it really is the space then the only
thing I can imagine is the space being some Unicode non-breaking space
char, you could check that with a hex editor...

<totally OT>
If it was the forced recompilation that fixed it, then something is
wrong with your dependency info; check your Makefile and/or run make
with verbose output to see what files it considers in its dependency
checks. But as changing the header finally resulted in a recompilation,
my guess would really be a desynced file time issue as I mentioned
earlier. Never use a source tree on NFS without NTP.
</totally OT>

Regards,
Malte