I haven't been able to reproduce the problem given your example. 

If you run doxygen with the -d commentcnv option it gives the result after replacing the /// comments with a /*..*/ internally. Maybe this helps you to debug the problem. I get:

/** \f{eqnarray*}
    N_{i,1}(x) & = & \left\{
      \begin{array}{ll}
        1 & \mbox{if $t_i \le x < t_{i+1}$ (inclusive on end knot)} \\
        0 & \mbox{otherwise}
      \end{array}
    \right. \\[8pt]
   
    N_{i,k}(x) & = &
      \frac{x - t_i}{t_{i+k-1} - t_{i}} N_{i,k-1}(x) +
      \frac{t_{i+k} - x}{t_{i+k} - t_{i+1}} N_{i+1,k-1}(x)
    \f} */

which is what I would expect.

Here's another example using doxygen 1.5.6 with the -d commentcnv command line parameter. The first formula still contains the "///" marks, whereas the second one is prefixed by "*/ /**". However, the third version seems to work just fine.

/// \brief Generate a symmetric Hann window.
///
/// Generate the \a iSize point symmetric Hann window \a ptArrayOut.
/// The window is defined as
/// \f[
///   w[n] = a - \left(1-a\right) \cos\left( 2 \pi \frac{n}{N-1} \right)
///          \quad \mathrm{with} \quad a = 0.50
/// \f]
    /// \f[
///   w[n] = a - \left(1-a\right) \cos\left( 2 \pi \frac{n}{N-1} \right)
///          \quad \mathrm{with} \quad a = 0.50
/// \f]
    /// \f[
    ///   w[n] = a - \left(1-a\right) \cos\left( 2 \pi \frac{n}{N-1} \right)
    ///          \quad \mathrm{with} \quad a = 0.50
    /// \f]
/// for \f$ 0 \leq n \leq N-1 \f$, otherwise zero.
template<class T> inline
T *genWindowHann(T *ptArrayOut, int iSize);

Maybe this helps a bit. It would be nice if someone could fix this bug before it gets a chance to celebrate its anniversiry. ;-)

Hey Dimitri, there is a new example in comment #2. I am thus reopening. However, feel free to ask any further question.

Created attachment 130749 [details]
sample code

Try running this through doxygen to confirm that "///" are incorporated
into the latex equation if the latex source spans several lines.

Hi,

  I can confirm this problem which has been annoying
me for a while. The bug must have crept in sometime
between version 1.4.4 and 1.4.7 (and it's still in 
1.5.8, the latest version I have installed).

 (I've just submitted sample code as an attachment
and that seems to have sent itself off without returning
me to this page, so I'm including the sample code again here.)
The key thing is that the "///" gets included as soon
as the latex source code spans more than one line.

Here is the html created when this file is processed
with doxygen version 1.3.9.1:

http://www.maths.manchester.ac.uk/~mheil/DUMP/doxygen_bug/html_version1.3.9.1/classJunk.html

and this is what it looks like when version 1.4.7
is used:

http://www.maths.manchester.ac.uk/~mheil/DUMP/doxygen_bug/html_version1.4.7/classJunk.html

Here's the example code:

//=============================================
/// This is a class that does stuff.
/// It solves the equation \f$ y'' + y = 0 \f$
/// or even
/// \f[
/// y'' + y' = f(x) 
/// \f]
/// or
/// \f[ y'' + y' = f(x) \f]
/// Pretty good or what?
//==============================================
class Junk
{
  public:
 
 Junk(){}

};





(In reply to comment #0)
> Please describe the problem:
> /// \f{eqnarray*}
> /// N_{i,1}(x) & = & \left\{
> ///   \begin{array}{ll}
> ///     1 & \mbox{if $t_i \le x < t_{i+1}$ (inclusive on end knot)} \\
> ///     0 & \mbox{otherwise}
> ///   \end{array}
> /// \right. \\[8pt]
> ///
> /// N_{i,k}(x) & = &
> ///   \frac{x - t_i}{t_{i+k-1} - t_{i}} N_{i,k-1}(x) +
> ///   \frac{t_{i+k} - x}{t_{i+k} - t_{i+1}} N_{i+1,k-1}(x)
> /// \f}
> 
> The generated formula for this fragment still contains the /// comment markup.
> Using /** */ style comments works.
> 
> I tried with a few different versions of doxygen. It worked in 1.4.2. It's been
> broken since 1.4.7 (1.4.6 screwed up Latex in a big way). Didn't try any other
> versions.
> 
> Steps to reproduce:
> 1. 
> 2. 
> 3. 
> 
> 
> Actual results:
> 
> 
> Expected results:
> 
> 
> Does this happen every time?
> Yes
> 
> Other information:
> 

Matthias, thank you very much for your input! Let's hope, Dimitri can have a look at it soon :)

Yes, the bugzilla is slow some times and doesn't return any pages. The actions, though, are performed most of the time.

Confirmed. Should be fixed in the next subversion update.

*** This bug has been marked as a duplicate of bug 336053 ***