Re: tic toc

[John W. Eaton]

On  5-Jan-2010, John Reed wrote:

| >>> "John W. Eaton" <address@hidden> 1/5/2010 10:19 AM >>>
| On  5-Jan-2010, John Reed wrote:
| 
| | Thank you for the explanation.  Since Octave is making system calls
| | to get the timing info it sounds like any bugs would be elsewhere in
| | my system rather than within Octave.
| | 
| | Some system information: 
| | 
| | running on XP service pack 2
| | 
| | it has intel core 2 duo processors (2.2 GHz)
| | 
| | and it uses visual C++ 2005.
| | 
| | Should I still report this issue to address@hidden even though the
| | bug appears to be in Windows?  If it would be useful I will go ahead
| | and report it.
| 
| OK, for Windows systems, we do some translation from the Windows
| system calls to POSIX interfaces.  In that case, maybe there is
| something we are not doing correctly that causes the error.  I also
| see that we could simplify the code in Octave somewhat by using some
| additional modules from gnulib, and that the way the gnulib functions
| do the job is different from what is currently done in Octave.  So
| first I think I will switch to using the gnulib modules.  But I have
| no way of testing these changes on a Windows system.  Once I make the
| changes, can you help by compiling Octave with the patch and seeing if
| it solves the problems for you?
| 
| sure i'd be happy to

I checked in the following changes:

  http://hg.savannah.gnu.org/hgweb/octave/rev/dcde57cb5778
  http://hg.savannah.gnu.org/hgweb/octave/rev/0b0bf1fd1ed7

These will not apply to the 3.2.x sources since they use gnulib
modules and that is a recent addition to the Octave sources.  You'll
need to build from the sources in the Mercurial archive to test the
changes.

jwe