[med-svn] [SCM] ray annotated tag, 0.0.2, created. 0.0.2

[Sébastien Boisvert]

The annotated tag, 0.0.2 has been created
        at  c6f45fa8db2855f2a7925b5f26c3a8da4b517609 (tag)
   tagging  ba7e0b53c09296af6a65afc9f4f68d304997ffba (commit)
 tagged by  Sébastien Boisvert
        on  Fri Feb 19 22:27:27 2010 +0000

- Shortlog ------------------------------------------------------------
* This version (0.0.2) works on Infiniband with as many cores as 1536. I removed any call to MPI_Barrier, because apparently, no synchronization is needed with Open-MPI 1.4.1.

Sébastien Boisvert (91):
      (null)
      Moving stuff around
      Loading a parameter: the directory name
      added a message limit
      Changes, test1, test2, and test3 are working properly
      MPI_Waitall is better
      * possibly proceed separatly for direct and reverse for edges.
      * possibly proceed separatly for direct and reverse for edges.
      * possibly proceed separatly for direct and reverse for edges.
      test1, test2, and SRA001125 file 1 load
      * free message memory
      commit 7938e7a2edaabf8ee0f55eb1b34e30f76e53c015
      test1 and test2 ok with coverage calculation
      Works ok.
      Sending code
      Added command for MPICH2
      commit 6686b74dbe4d2e6ad6577bd1d978602b5d4b77cc
      commit 72db23dd80245fe0706462eda62e138ff2ce94b4
      commit 3c9cb7ae8bcba1e7c0df4a136cdb90536bf11065
      commit 1e3ad1ba3bcb24442b441740b1d42d290fb88c0c
      commit 639ecd60b249491e3523c7d596aa72461297a887
      commit 107f7de8cc18564fe23e7332dd799451e10f067e
      commit 800b776a78e2fa752914917fb5174dfa4381fdc6
      * the seeds are now computed correctly. (need to test stuff in TODO.txt)
      * Fixed problem with Finalize and MPI_Barrier.
      * Fixed a bug in building seeds.
      * will fix these later or never (they are not that important.):
      Seems ok to me:
      * Reads get attached now.
      * Ray returns immediatly if given an invalid configuration file.
      * added version in stdout
      commit 370f620230c9676e0ca8d5dce77ae8f1b33d4cda
      * Show progress when attaching reads.
      * show progress for vertex seeding.
      No message, just commiting stuff.
      Done: now I have 2*n seeds! yay.
      Reduced communication:
      * Fixed a SEG FAULT when sending large sequences.
      Updated TODO .
      This behavior is correct:
      * The coverage of a vertex is now stored as a 'char', with a maximum of 255.
      * Edges are stored as a 'char'.
      * fixed compilation warnings.
      * removed cout's
      The memory usage is lowered, now each vertex is a (uint64_t,char,char,bool)
      Summary of today's work:
      commit ae3f25682210aca70f34272ce21ddeccd18152b9
      Fixed:
      * The extension is now done in both direction.
      fixed compilation warnings
      * I added constants for allocator's chunk sizes.
      * only keep seeds with at least 100 nucleotides.
      * removed commented code lines in the code.
      * FIXED: random mismatches plague seeding?
      * remove Edge class (it was replaced by ReadAnnotation)
      FIXED:
      Added this feature:
      Updated TODO.
      Updating TODOs
      Directions are copied.
      * Restored extension.
      Implemented:
      Fixed misassembly.
      Setting the PERIOD for MPI_Barrier back to 100
      Removed paired-end Class.
      Added the direction class.
      * removed the line break before the Minimum Coverage
      The fusion has begun.
      WORKS-FOR-ME:
      Fixed:
      Fixed:
      Fixed compilation errors.
      Fixed:
      Fix in README.
      The next milestone:
      Updated TODO.
      working on a bug
      good .
      * Removed the DUMMY message.
      Currently investigating the hang.
      The hang is caused by the btl sm.
      Removing ther Makefile for colosse.
      Testing speed limitation.
      Speed regulator works on Llactis with 8 MPI processes.
      * Removed all MPI_Barrier calls from the program (they are broken with shared memory for
      Implemented:
      * added progress when attaching reads.
      SRA001125 preliminary results:
      * Remove calls to Get_processor_name in code.
      * Removed time.h and sys/time.h from includes.
      * This version (0.0.2) works on Infiniband with as many cores as 1536.

-----------------------------------------------------------------------

-- 
Packaging of Ray in Debian