[forwarded from http://bugs.debian.org/136028
 forwarded by email as well, but no trace in the BTS]

I think I've found a couple of bugs. The first is with function parameter 
detection. The new doxygen version generates many warning regarding missing
variable in function definitions. For example for this function
/**
 * @fn aggiorna_par_rx_can(unsigned int,unsigned char*,unsigned char)
 * @param id_msg msg identifier
 * @param vect_rx answ. vect
 * @param length_mess message lenght
*/
void aggiorna_par_rx_can(unsigned int id_msg,
    unsigned char* vect_rx,
    unsigned char length_mess)
{
...
}

I get the following errors:

can.c:449: Warning: argument `id_msg' of
command @param is not found in the argument list of
aggiorna_par_rx_can(unsigned int, unsigned char *vect_rx, unsigned char)
can.c:449: Warning: argument
`length_mess' of command @param is not found in the argument list of
aggiorna_par_rx_can(unsigned int, unsigned char *vect_rx, unsigned char)
adb800/can.c:449: Warning: The following
parameters of aggiorna_par_rx_can(unsigned int, unsigned char *vect_rx,
unsigned char) are not documented:
  parameter int
  parameter char

But it generates the right output. I'm sure the problem is in the latest
doxygen version. 

The second problem is with the automatic link generation. I noticed that
when I define a group that assemble together functions from different
files, many anchor links in each files, refer to the local file instead
of the group file. For example if I have two file with functions:
a.c
	foo()
	bar()
b.c
	barfoo()
	foobar()

and I define a group "c" that assembles foo() and foobar(), I'll have
the anchor links in a.c point to a_c.html#a1 instead of group_c.html#a1.
This is true for in-functions anchor links (that is, the anchor defined
in the function code) but not for the header links (that is, links defined 
at the beginning of a_c.html that groups functions, variables, definitions, 
etc).

thank you for your help,
grazie dell'aiuto,
		pietro

Luca Filipozzi wrote:

Hi Dimitri,

the following bugs have been reported against Doxygen.  Can you let me
know if you consider them to be valid and whether you plan to fix them?

Thanks very much,

Luca

--------------------------------------------------------------------------------
#136028: doxygen: warning and link problems - Part A: (I think this is fixed.)
  I think I've found a couple of bugs. The first is with function parameter 
  detection. The new doxygen version generates many warning regarding missing
  variable in function definitions. For example for this function
  /**
   * @fn aggiorna_par_rx_can(unsigned int,unsigned char*,unsigned char)
   * @param id_msg msg identifier
   * @param vect_rx answ. vect
   * @param length_mess message lenght
  */
  void aggiorna_par_rx_can(unsigned int id_msg,
      unsigned char* vect_rx,
      unsigned char length_mess)
  {
  ...
  }

  I get the following errors:

  can.c:449: Warning: argument `id_msg' of
  command @param is not found in the argument list of
  aggiorna_par_rx_can(unsigned int, unsigned char *vect_rx, unsigned char)
  can.c:449: Warning: argument
  `length_mess' of command @param is not found in the argument list of
  aggiorna_par_rx_can(unsigned int, unsigned char *vect_rx, unsigned char)
  adb800/can.c:449: Warning: The following
  parameters of aggiorna_par_rx_can(unsigned int, unsigned char *vect_rx,
  unsigned char) are not documented:
    parameter int
    parameter char

  But it generates the right output. I'm sure the problem is in the latest
  doxygen version. 
  I think I've found a couple of bugs. The first is with function parameter 
  detection. The new doxygen version generates many warning regarding missing
  variable in function definitions. For example for this function
  /**
   * @fn aggiorna_par_rx_can(unsigned int,unsigned char*,unsigned char)
   * @param id_msg msg identifier
   * @param vect_rx answ. vect
   * @param length_mess message lenght
  */
  void aggiorna_par_rx_can(unsigned int id_msg,
      unsigned char* vect_rx,
      unsigned char length_mess)
  {
  ...
  }

  I get the following errors:

  can.c:449: Warning: argument `id_msg' of
  command @param is not found in the argument list of
  aggiorna_par_rx_can(unsigned int, unsigned char *vect_rx, unsigned char)
  can.c:449: Warning: argument
  `length_mess' of command @param is not found in the argument list of
  aggiorna_par_rx_can(unsigned int, unsigned char *vect_rx, unsigned char)
  adb800/can.c:449: Warning: The following
  parameters of aggiorna_par_rx_can(unsigned int, unsigned char *vect_rx,
  unsigned char) are not documented:
    parameter int
    parameter char

  But it generates the right output. I'm sure the problem is in the latest
  doxygen version. 

--------------------------------------------------------------------------------
#136028: doxygen: warning and link problems - Part B:
  The second problem is with the automatic link generation. I noticed that
  when I define a group that assemble together functions from different
  files, many anchor links in each files, refer to the local file instead
  of the group file. For example if I have two file with functions:
  a.c
    foo()
    bar()
  b.c
    barfoo()
    foobar()

  and I define a group "c" that assembles foo() and foobar(), I'll have
  the anchor links in a.c point to a_c.html#a1 instead of group_c.html#a1.
  This is true for in-functions anchor links (that is, the anchor defined
  in the function code) but not for the header links (that is, links defined 
  at the beginning of a_c.html that groups functions, variables, definitions, 
  etc).