============================ Pre-release of CALC/SOLVE ======================== (The last released version of CALC/SOLVE: 2005.01.28) === Headlines === 1. REPA was completely rewritten from scratches. Documentation is available at $MK5_ROOT/help/repa_01.txt and $MK5_ROOT/help/repa_02.txt READ ATTENTIVELY DOCUMENTATION THREE TIMES BEFORE THE FIRST USE!! (2004.12.16) 2. Support of two new data types was added: Narrow-band group delay at X-band Narrow-band group delay at S-band $MK5_ROOT/libs/cutil/datyp_show.f $MK5_ROOT/libs/cutil/datyp_inq.f $MK5_ROOT/libs/cutil/make_oc.f (2004.12.16) 3. New feature was added to REPA: ability to display fringe plots. Refer REPA documentation for more details. $MK5_ROOT/libs/cutil/flyby_init.f $MK5_ROOT/libs/cutil/init_glb_cm4.f $MK5_ROOT/progs/solve/newdb/n_adds.f $MK5_ROOT/progs/solve/newdb/n_puts.f $MK5_ROOT/progs/solve/sdbh/geth1.f $MK5_ROOT/progs/solve/sdbh/geto2.f $MK5_ROOT/progs/solve/sdbh/blkcl.f $MK5_ROOT/progs/solve/repa/mk4plot.f $MK5_ROOT/progs/solve/repa/repa_df_init.f $MK5_ROOT/progs/solve/repa/repa_setdiagi.f $MK5_ROOT/progs/solve/repa/repa_load_obs.f $MK5_ROOT/progs/solve/repa/repa_df_fp.f $MK5_ROOT/progs/solve/cres/cremu.f (2005.01.28) === Improvements === 1. DiaGi and Multu_DiaGi were significantly improved. Read the bottom of $MK5_ROOT/help/diagi_1.htp for more details. $MK5_ROOT/libs/diagi/diagi_inq.f $MK5_ROOT/libs/diagi/diagi_res.f $MK5_ROOT/libs/diagi/diagi_hlp.f $MK5_ROOT/libs/diagi/diagi_mul.f $MK5_ROOT/libs/diagi/diagi_met.f $MK5_ROOT/libs/diagi/diagi_dra.f $MK5_ROOT/libs/diagi/diagi_chf.f $MK5_ROOT/libs/diagi/diagi_sav.f $MK5_ROOT/libs/diagi/diagi_prn.f $MK5_ROOT/libs/diagi/diagi.f $MK5_ROOT/libs/diagi/multi_diagi.f $MK5_ROOT/libs/pgplot_plus/xwdriv_plus.c (2004.12.15) 2. PWXCB was ported under Linux (David Gordon). (2004.01.24) === Bug fixes === 1. Fixed bugs in MUL_MM_TI_I and MUL_MM_TI_T . At small dimensions extra loop cycles were executed. $MK5_ROOT/libs/matvec/mul_mm_ti_i.f $MK5_ROOT/libs/matvec/mul_mm_tt_i.f (2004.11.15) 2. Consolidated setting modes and flags in low level routines for file opening. The old versions used hard-coded values which may be different at different operating systems. $MK5_ROOT/libs/pet_util/get_system_constant.c $MK5_ROOT/libs/pet_util/binio.f $MK5_ROOT/libs/fclib/fciv_open.c $MK5_ROOT/libs/cutil/bin_create.f $MK5_ROOT/libs/cutil/bin_open.f $MK5_ROOT/libs/cutil/bin_open_rd.f $MK5_ROOT/libs/cutil/make_sema.f $MK5_ROOT/libs/cutil/basfe_file.f $MK5_ROOT/libs/cutil/crate_arcf.f $MK5_ROOT/libs/b3d/fast.f $MK5_ROOT/libs/b3d/fast_glo.f (2004.11.15) 3. Fixed a sleeping bug in chain4. Ihash.f could not handle a return from the crcc hash routine equal to -32768, because ihash then found the absolute value but put it in an integer*2 variable, which was not equipped to handle a value of +32768. (On HP-UX, the value remained as -32768, which is an invalid hashfile record and later caused problems in readhash.) To get around this, an integer*4 variable is now used. $MK5_ROOT/libs/chain4/ihash.f (2004.11.15) [kdb] 4. Fixed another sleeping bug in chain4. If there was an error reading the hashfile, and the user wanted to stop trying to read it, the code failed to stop trying. $MK5_ROOT/libs/chain4/readhash.f (2004.11.15) [kdb] 5. Fixed sskedh: the previous version interpreted the second word in $OBS record as the total number of observations. Looks like it was not the case. $MK5_ROOT/progs/solve/sskedh/gethd.f $MK5_ROOT/progs/solve/sskedh/rd_sked_source_list.f (2004.11.15) 6. Fixed bug in batch reported by Dorothee Fisher: Batch may fail in restarting with message "BATCH: ERROR 116 Error 31488 opening" Reason: uninitialized error code variable. $MK5_ROOT/progs/solve/batch/eof_spool.f (2004.11.16) 7. Fixed bug in batch Solve: the previous version incorrectly processed global clock polynomial flags, because the logic is completely rotten down, and is the data structure are fabulously inadequate to algorithms. $MK5_ROOT/progs/solve/batch/ofsts_mult_ref.f (2004.11.26) 8. Fixed a bug: in many places Solve read with format *, and relied on IOSTAT variable. This does not always work, since this specification is very flexible and accepts some non-numerical values. $MK5_ROOT/progs/solve/batch/gflags.f $MK5_ROOT/progs/solve/batch/garc.f $MK5_ROOT/progs/solve/batch/gconst.f $MK5_ROOT/progs/solve/batch/shfeop.f $MK5_ROOT/progs/solve/batch/find_weights.f $MK5_ROOT/progs/solve/batch/prces.f $MK5_ROOT/progs/solve/batch/ctrls.f $MK5_ROOT/progs/solve/batch/batch.f (2004.11.29) 9. Fixed a bug: hsort_key relied on arithmetic operations with character data. This cannot work the same at different architectures. Removed this routine. $MK5_ROOT/progs/solve/hausr/outwithitall.f (2004.11.30) 10. Recent changs awoke a sleeping bug: EOP constraints bit logic was wrong: if EOPMOD = 'N', then EOP constraint bit should be lifted. $MK5_ROOT/progs/solve/batch/arcset.f (2004.12.02) 11. Impoved appearance of the error message: since the command string was not initialized, garbage may be printed. 12. Fixed two errors in dbedit: the index at SOB array was ommitted which resulted in crash. The old version did not complain at the wrong keyword in $INPUT_FILTER section. The new version abnormally terminates when it finds unsupported value. $MK5_ROOT/progs/dbedit/read_ddbuf.f $MK5_ROOT/progs/dbedit/get_filter.f (2004.12.27) 13. Fixed eop_alignment: IERS C04 format was changed in 2005. $MK5_ROOT/utils/eop_alignment/rd_iers_c04.f (2005.01.13) 14. Fixed error in repa. The previous version did not handle correctly the situation when ambiguities at S-band were resolved. $MK5_ROOT/progs/solve/repa/repa_com.f $MK5_ROOT/progs/solve/repa/repa_updamb.f (2005.01.14) 15. Fixed several glicthes occurred during compilation reported by Wolfgang Schwegmann. $MK5_ROOT/libs/matvec/invsf.f $MK5_ROOT/testware/matvec_test/matvec_test.f (2005.01.20) 16. Fixed several errors in MDLPL related to the fact that ICODE variable for MDLPL_DIAGI was not defined. $MK5_ROOT/progs/solve/make_psfplot.f $MK5_ROOT/progs/solve/mdlpl_plus.f --------------------------------------------- Leonid Petrov 2005.01.28_20:52:01