Commits on Jan 15, 2008
  1. Changed the default logconf file location from /tmp/.vw_logconf to ~/…

    ….vw_logconf. Also updated the release notes for a VW 2.0alpha4 release.
    broxtronix committed Jan 15, 2008
  2. Started working through the GPU code to bring it in line with normal …

    …VW coding conventions.
    broxtronix committed Jan 15, 2008
  3. Added a config flag for changing the target architecture (to support …

    …64-bit compiliation on MacOS). Added better config output for selecting the number of threads.
    broxtronix committed Jan 15, 2008
Commits on Jan 14, 2008
  1. Fixed an uninitialized variable warning, and added an assertion for

    the assumption that made it not actually be a problem.
    Matthew Hancher committed Jan 14, 2008
  2. Fixed a tricky edge condition of sorts in ReflectEdgeExtension,

    and added a regression test.  My first commit in many weeks!
    Matthew Hancher committed Jan 14, 2008
Commits on Jan 7, 2008
Commits on Jan 5, 2008
  1. Fixed the ThreadPool unit test to work on the linux machine. Fixed a …

    …few more compiler warnings.
    broxtronix committed Jan 5, 2008
Commits on Jan 4, 2008
  1. Moved the KDTree from Interest point to math module and cleaned up so…

    …me namespace issues (it had a "using namespace boost;" at the top of it that was likely to lead to trouble down the road.
    broxtronix committed Jan 4, 2008
  2. Checking in a new chapter of the Vision Workbench on the Core Module.…

    … Topics covered include Exceptions, Caching, and the System Log. Also checking in some changes to the doxygen docs for Cache and Exception.
    broxtronix committed Jan 4, 2008
  3. Fixed numerous small compiler warnings. Only one (in GloabalToneMap.h…

    …) was of any consequence.
    broxtronix committed Jan 4, 2008
Commits on Jan 2, 2008
  1. Checking in a fix for a small typo in ThreadPool.h that inadvertantly…

    … got checked in a sec ago.
    broxtronix committed Jan 2, 2008
  2. Adding some debugging messages in the CorrelatorView, ImageResourceVi…

    …ew, and EdgeExtensionView
    broxtronix committed Jan 2, 2008
Commits on Jan 1, 2008
  1. Added proper handling of std::endl and std::flush to log streams. Als…

    …o added some thread safety (a mutex) to the ImageResourceView so that only one thread can call read() on a DiskImageResource at a time. Added a multi-threaded version of the interest point detection algorithm, which seems to work.
    broxtronix committed Jan 1, 2008
Commits on Dec 31, 2007
  1. Changed calls to vw_out() throughout the vision workbench to include …

    …log_namespace values so that logging can be more carefully controlled. Added some thread safety to the SystemLog class. Finished the implementation of the code that periodically re-reads the /tmp/.vw_logconf file to see if there are changes to what should be logged. I will send out email later tonight or tomorrow describing the features of the new logging system in detail. TODO: 1) Right now the .vw_logconf file blows away any log settings you have made through the API. Eventually this should be fixed so that both are effective at changing log settings. 2) We still don't handle std::flush and std::endl correctly in all cases. I mystified. In the meantime, the console may not always refresh where these stream modifiers appear at the end of a line of output.
    broxtronix committed Dec 31, 2007
Commits on Dec 28, 2007
  1. Fixed a few bugs associated with flushing the log stream on a cairrag…

    …e return (mostly for TerminalProgressCallback). Fixed a small bug in the cxxtest::TestSuite that appears to be a conflict with boost. Not sure why it hasn't manifested itself before...
    broxtronix committed Dec 28, 2007
  2. Fixed a small typo

    broxtronix committed Dec 28, 2007
  3. First switch-over to the new logging system. vw_out() is now a thin

    wrapper around the default log (to the console), and much of the
    infrastructure is there to log to files (or really any type of ostream).
    Checking the code in now and moving to octo to work on LogRuleSets.
    broxtronix committed Dec 28, 2007
Commits on Dec 27, 2007
  1. First cut at a new and improved logging facility. This compiles with …

    …some basic unit tests, but it's not finished yet. I'm just checking this in for now so that the work-in-progress is backed up. This all lives in and Log.h for now, so it shouldn't affect anything else.
    broxtronix committed Dec 27, 2007
Commits on Dec 25, 2007
  1. Added a static method, Thread::id(), which returns a globally unique

    integer identifier for the thread from which the method is called.
    This is generally handy, and it will be first used to enable the new
    logging code to keep track of threads and to keep their logging
    streams internally separate from one another.
    broxtronix committed Dec 25, 2007
Commits on Dec 23, 2007
  1. One last fix for automake 1.10. For some reason it invokes the sh bui…

    …lt-in echo instead of /bin/echo. Sh echo does not have the -n flag. Prior to this fix, the default enable state for packages was being ignored.
    broxtronix committed Dec 23, 2007
  2. More changes to the automake files to support automake 1.10. Added qu…

    …otations around variable names in calls to the test command.
    broxtronix committed Dec 23, 2007
  3. Checking in some changes for compatibility with automake 1.10 (which …

    …is installed by default on OSX 10.5)
    broxtronix committed Dec 23, 2007
Commits on Dec 19, 2007
  1. Correct Luv from L value constructor.

    Xavier Bouyssounouse committed Dec 19, 2007
  2. Add Luv color space.

    Xavier Bouyssounouse committed Dec 19, 2007
Commits on Dec 18, 2007
  1. fixing ctor parameter type for slog_width from int to float as

    m_slog_width is float, too. - Triggered a type-mismatch warning in tp, too.
    Hans Utz committed Dec 18, 2007
  2. - trying to fix LAPACK discovery problems. the logic should be more s…

    towards autoconf bugs. but i only tested on boink.
    - trying to fix QHULL discovery problems. adding the missing headers
    to the test-macro helps. but now the question is, why it worked in 
    the first place.
    Hans Utz committed Dec 18, 2007
  3. Added a caching feature for GDAL File I/O that allows many image hand…

    …les to be open (or close, but cached) simultaneously.
    broxtronix committed Dec 18, 2007
Commits on Dec 11, 2007
  1. Added an optimization to avoid calling proj.4 when the source and des…

    …t projection are the same in GeoTransform
    broxtronix committed Dec 11, 2007