changelog

用于2维的射线追踪

----------------------------------------------------------------------------
1.0 -  
----------------------------------------------------------------------------
* mardi 15 mai 2007, 09:51:48 (UTC+0200)
 - ok. Stable enough now, tested on many platforms to go for 1.0 !

0.18 -----------------------------------------------------------------------
Objectives : * code stabilization 

* Dim mai 13 18:47:57 CEST 2007
  - add more timers, and specifically to time merge.c:send_recv_cell_info()

* mar jun 28 12:26:59 CEST 2005 - marc
  - compilation fixed when using ZLIB

* Sat Jun 18 10:51:42 CEST 2005 - marc
  - add --tmp directory option 
  
* Fri Jun 17 18:31:33 CEST 2005 - marc
  - shown mem usage when starting merge, after buffer importation, at merge end
  - back to use MPI_Waitany rather than MPI_Waitall in phase merge

* Fri Jun 10 16:16:01 CEST 2005 - marc
  - sync mpi SCATTER mode whith master/save version

* sam avr 23 11:18:43 CEST 2005 - marc
  - use geo_middleb rather than geo_middle, as geo_middleb 
    always returns the center of the shorthest arc between p1 and p2

* dim avr 17 13:14:45 CEST 2005 - marc
  - version 0.18
  - add an interpolated node when ray segment is too long, so the ray 
    could miss a cell. This bug corection fixes :
    	. a lot of "intersection bug"
    	. face hit corruption
	. error on ray length

0.17 -----------------------------------------------------------------------
Objectives : * merger can produce sparse and sco files
             * implements more sofisticated ray filter, based
	       on epicentral dist, time residual, nb of ray in the bundle
	     * manage enhanced phase code
	       
* mer avr  6 16:06:04 CEST 2005 - marc
  - can manage enhanced phase code.

* Mon Mar  7 14:18:57 CET 2005 - marc
  - sync with new sparse fonction (lsqrsolve)

* ven mar  4 19:15:04 CET 2005 - marc
  - added filter.{c,h}
  - starting more sofisticated ray filter implementation

* jeu mar  3 15:55:52 CET 2005 - marc
  - version 0.17
  - merger can produce sco files, also updates the .xml
    accordingly
  - merger can produce sparse file
  - add --sparse and --sco options
    
	
0.16 -----------------------------------------------------------------------
Objectives : * ultra super mega dataset size
	     * added raybundle
	     * added filter on residual time value

* mer mar  2 00:52:18 CET 2005 - marc
  - added filter on residual time value '--res-filter' option
  - version 0.16.3

* dim f関 27 00:03:43 CET 2005 - marc
  - sequential version fixed for sparse and res files

* dim f関 20 10:30:24 CET 2005 - marc
  - version 0.16.2
  - [mpi/master slave] renumbering res file fixed

* ven f関 11 15:29:20 CET 2005 - marc
  - full working ray bundle
  - debug mode disable progress bar use

* mer f関  9 14:08:33 CET 2005 - marc
  - added '-s' option to write a sco file  

* mar f関  8 16:25:29 CET 2005 - marc
  - ligth mesh to real mesh coordinates convertion (sco file)

* lun f関  7 17:50:39 CET 2005 - marc
  - ray bundle completed

* ven f関  4 15:44:38 CET 2005 - marc
  - ray bundle basicaly working
  - installation of documentation png files

* jeu f関  3 18:14:18 CET 2005 - marc
  - migration of which_layer(), point2cell() to libmesh
  - add raybundle

* jeu jan 27 17:44:57 CET 2005 - marc
  - version 0.16.1
  - get_mem_usage() uses mallinfo()
  - configure checks for mallinfo

* lun jan 17 11:52:27 CET 2005 - marc
  - add get_mem_usage() using /proc for linux, other unices should
    use getrusage.

* lun jan 10 17:10:50 CET 2005 - marc
  - more layer argument check

* mer d閏  8 16:57:34 CET 2004 - marc
  - merger can import users specified layers using -b and -e flags

* lun d閏  6 18:00:35 CET 2004 - marc
  - merger synchro with last ray2mesh release

* ven d閏  3 17:46:49 CET 2004 - marc
  - sequential version

* mar nov 30 17:12:48 CET 2004 - marc
  - version 0.16
  - one can use -L or --limit option to specify the maximum memory size to use
    approximatively. Several r2m files are then created.

0.15 -----------------------------------------------------------------------
   
* lun nov 22 10:58:50 CET 2004 - marc
  - version 0.15
  - sync with mesh-1.0pre3
  - fixed memleak in r2m_cellface_orientation_get, points where not desallocated

0.14 -----------------------------------------------------------------------
objectives : ACH tele-seismic mode

* mar nov 16 17:44:50 CET 2004 - marc
  - version 0.14.1
  - rewrite ray2mesh as ray2mesh2 : one probleme remains, 
   the exact entering point is not computed in ACH mode

* mer nov 10 15:48:20 CET 2004  - marc
  - remove previous patch, and add a file with 
    event_id ray_id propagation_time
* mer nov  3 18:35:42 CET 2004 - marc
  - for global tomo, put all the ray length before entering into the mesh
    in the first cell. And put all remaining length in the last visited 
    cell of the mesh


0.13 -----------------------------------------------------------------------
objectives : a full working mpi version with XML files construction.
	     ray re-numbering should be written somewhere in the 
	     RES and SPARSE section of the XML.

* ven oct 22 11:12:18 CEST 2004 - marc
  - fixed header for res and sparse files
  - do not stop even a domain has only empty cells
* sam oct 16 11:40:14 CEST 2004 - marc
  - clean-up code
* ven oct 15 18:47:01 CEST 2004 - marc
  - remove duplicate code
  - used mesh2xml

0.12 -----------------------------------------------------------------------
objectives : direct problem 

* jeu oct  7 11:33:53 CEST 2004 - marc
  - uniformize r2M and sco file header
* mer oct  6 11:15:03 CEST 2004 - marc
  - sparse matrix format change (add rayid)
  - add header to sparse file
* lun oct  4 14:11:07 CEST 2004 - marc
  - move linearize_cell_id() from ray2mesh to libmesh
* mer sep  8 11:57:31 CEST 2004 - marc
  - create the files A and dT where A.dx=dT [done]

0.11.5 ---------------------------------------------------------------------
* dim mai  2 23:45:56 CEST 2004 - genaud
  - coformance to new xml format : added a dump_enriched_xml() to 
    generate a <data> section (for seq/scatter/ms versions).
0.11.4 ---------------------------------------------------------------------
* mar avr 27 12:28:50 CEST 2004 - genaud
  packages + output fixes (ray2mesh_static must be asked by configure --enable-static)
  man page ray2mesh.8 added
0.11.3 ---------------------------------------------------------------------
* mar mar 16 17:39:57 CET 2004 - genaud
  make ray2mesh adapted to versatile libiasp (C or Fortran) with
  --enable-fortran and --with-model-name=x in configure
0.11.2 ---------------------------------------------------------------------
Tested on 1999-2000.ray (iterative mode). Better outputs.
0.11.1 ---------------------------------------------------------------------
* jeu mar  4 15:47:43 CET 2004 - genaud
  - bugfix : no intersection point found => insert an arbitrary point instead
    segfaulting
* ven f関 27 14:21:44 CET 2004 - genaud
  - ray2mesh() completely rewritten. Made EXACT_LENGTH the standard and 
    removed approx. length. Checked against mem-leaks ok.
0.11.0 ---------------------------------------------------------------------
objectives : API re-organization. All cellinfo management has moved to
libmesh. Requires libmesh >= 0.8.0
* mer f関 25 18:03:25 CET 2004 - genaud
  - use intersection points in in/out cell points when --enable-el is on
* sam f関 21 11:50:45 CET 2004 - genaud
  - moved allmost all cell_info_t management to libmesh (removed
    output.c and feed.c)
  - rewrote functions calls to conform API mesh_cellinfo_<object>_<action>()
             
0.10.5---------------------------------------------------------------------
objectives : improvements 
* ven f関 20 12:34:18 CET 2004 - genaud
  - rayids are unique but also consecutive in the master/slave version
  - debug trace reworked
0.10.4---------------------------------------------------------------------
objectives : unique rayid generation for gimbos 
* sam f関 14 15:45:38 CET 2004 - genaud
  - generation of a unique rayid in parallel implementations
  - bug fix (bug introduced in 0.10.3) : --r2moutput did not work in master-slave impl.
0.10.3 ---------------------------------------------------------------------
objectives : code review and re-organization following --limit 
* mer jan 28 16:42:52 CET 2004 - genaud
 - added a 'merger' binary :
   merger.c : basic new code to merge overlapping r2m files
 - ray2mesh.c splitted in two (ray2mesh.c + feed.c)
 - moved dup_coord_geo() to raydescartes (coord.c)
 - fixed output filenames generated with --r2moutput
 - make --outputfile and --r2moutput mandatory with --limit
* lun jan 26 16:29:19 CET 2004 - genaud
 - complete rewrite of the output file opening scheme.
 - moved  configure.in -> configure.ac + fixed bug for MPI detection
0.10.2 ---------------------------------------------------------------------
objectives : implement --limit option for limited memory comsumption
* ven jan 23 14:12:29 CET 2004 - caprini
 - implemented -L|--limit x : dumps cell_info to disk when memory is going
  beyond x MB
0.10.1 ---------------------------------------------------------------------
objectives : computation of exact lengths (for ray lengths)
* mar jan 13 21:56:47 CET 2004 - genaud
 - memory leaks appeared in the new length code removed
* mar jan  6 15:05:41 CET 2004 - caprini
 - raymaths.c : const EPS_10_6, - EPS_10_6 < mu < 1 + EPS_10_6 
* mer d閏 24 17:54:12 CET 2003 - genaud
 - feed_single_cell_hit_info() rewritten. Previous was buggy.
* mer d閏 24 01:28:08 CET 2003 - marc
 - hit_face_score() seems to work - no more intersection problem detected
* mar d閏 23 03:17:47 CET 2003 - marc
 - hit_face_score() reworked
* sam d閏 20 19:15:38 CET 2003 - marc
 - add files to generate html/latex documentations
 - re-format some files to conforme to doxygen
* ven d閏 19 16:03:51 CET 2003 - caprini
 - intersections of rays with plane faces, spherical faces, conic faces to
   store the exact length of rays in cells. Define EXACT_LENGTH to activate
   this part of code.

0.10.0 ---------------------------------------------------------------------
objectives : 
	- generation of 2 kinds of files :
	  o .sco files for score file,
	    format = [x,y,z] s1 s2 ... sn
	  o .r2m files to show all informations

* sam nov 22 18:25:13 CET 2003 - genaud
 - add --filtered|-f option. Followed by a filename, indicates to store ray
   data items that are traced RAY_TRACE_OK
 - DEBUG_INTERSECT is replaced by EXACT_LENGTH. Define this to build
   intersection tests
* mer nov 19 11:19:11 CET 2003 - marc
 - bug fixed when importing r2m domain, items were not imported  
* ven nov 14 19:56:16 CET 2003 - marc
 - add DEBUG_INTERSECT flags to set/remove verbose intersection output 
* dim sep 21 09:09:46 CEST 2003 - marc
 - change get_raydata to accept broken ray line, just ignore it, with a warning
* mar sep  9 17:04:13 CEST 2003 - marc
 - ray2mesh conforms to r2m and sco file format needed by mesh
* mar sep  2 14:56:30 CEST 2003 - marc
 - reflect RAYCODE_NOT_MATCH change in raydescartes, 
   ray computed/rejected statistics are ok.
* dim ao