Rotate for json files. Fixes in nfde_rotate #188
Conversation
adrianarce-elemwave
force-pushed
the
feature/rotate-json
branch
from
5dec29f to
321e8aa
Compare
adrianarce-elemwave
force-pushed
the
feature/rotate-json
branch
from
847a93d to
68a330b
Compare
test/rotate/CMakeLists.txt
Outdated
There was a problem hiding this comment.
I think it is excessive to create so many files to test this function. Could you consider to put them in a single file (as separate tests) and include it maybe within test/system/ folder?
| @@ -40,15 +40,17 @@ END SUBROUTINE nfde_rotate | |||
| SUBROUTINE rotate_generateSpaceSteps (this, mpidir) | |||
There was a problem hiding this comment.
why is the name of these functions rotate_generateWhatever? I understand these are rotating structures but there is no generation, which is made during parsing.
| newrotate=.false. !!ojo tocar luego | ||
| #ifdef CompileWithSMBJSON | ||
| newrotate=.true. |
There was a problem hiding this comment.
If we are going to always use the new rotate, then we should remove this flag.
adrianarce-elemwave
force-pushed
the
feature/rotate-json
branch
5 times, most recently
from
c9f099b to
969c22b
Compare
src_main_pub/semba_fdtd.F90
Outdated
| DO | ||
| READ (UNIT_EF, '(A)', end=1010) l_aux | ||
| rawFileInfo%numero = rawFileInfo%numero + 1 | ||
| IF (len_trim (adjustl(l_aux))>=BUFSIZE) then |
There was a problem hiding this comment.
When json files are modified using the json python module, they become oneliners. Many tests change minor things in the the json files to be able to tests several cases from a single test (chaning the materialId in the material association), or changing the number of steps, to run shorter tests. When those modified json reach this point, they are read a single line whose size is larger than BUFSIZE.
There was a problem hiding this comment.
Attempted fix:
- first, count the number of lines in the file
- if > 1, read the same way
- if == 1 and the file is *json, read chunks of the line (which has a large but a priori undefined size) with size 1024 (bufsize), and make newlines every time a "}" is found
There was a problem hiding this comment.
Changed: it is irrelevant that lines end in some particular separator. Reading up to every "}" risks not reading the content between the separator and the end of the buffer. NOW: split the json oneliner into chuncks of 1024 characters, each one will be a line
No description provided.