open TELEMAC-MASCARET The mathematically superior suite of solvers

2 problems - both addressed in telemac3d.f

All changes occur around line 2613 V7r1p0.

1) the restart mode call to bief_desimp should be callable at the termination of computation regardless of the graphic output values. Only required check is that

LT.EQ.NIT.AND.RESTART_MODE

else you have to ensure that it is also matched to the graphic output periods and intervals test. Move it outside the test

IF(MOD(LT,GRAPRD).EQ.0.AND.LT.GE.GRADEB) THEN

2) the call to bief_desimp needs to set the timestep (data record number) to zero. This is interpreted as the record number in add_data - if greater than 0 for the first write it seeks to beyond the end of the file to write its record - leaving the file unreadable. For the restart mode file only one data record is written - hence record=0.

I thought that this was related to the Int8 requirement for writing large files previously submitted but it isnt.

Attached is also a simple python script that can check the coherence of a selafin file. This can be use for tracking down file write problems. Works on all the selafin files that i have been playing with.

Note this may also apply to telemac2d!

Cheers
John

Thought I better put in the call to bief_desimp that works around telemac3d Line 2614 (note this call has been moved outside the result write code)

IF(LT.EQ.NIT.AND.RESTART_MODE
CALL BIEF_DESIMP(T3D_FILES(T3DRST)%FMT,VARSO3,NPOIN3,
& T3D_FILES(T3DRST)%LU,BINRES,AT,0,
! **** changed to zero ^
& 1,NIT,
& SOREST,SORIS3,MAXVA3,TEXT3,1,NIT)
ENDIF

Cheers
john

Hello,

Well done ! This was actually identified and will be corrected in the revision R1 of version 7.1, which will be issued very soon.

With best regards,

Jean-Michel Hervouet