8832c04dd0
more sanity checks
...
number of arguments for values per system needs to match the number of
systems that are defined
2018-12-21 11:52:23 +01:00
a1ff380ef4
slightly better aligned debug output
2018-12-20 17:14:55 -05:00
b2062f2a12
label were stored including [] and comments
2018-12-17 16:11:01 +01:00
72304638f9
PGI fails if optional argument is not present
2018-12-06 00:43:32 +01:00
29e55d20fb
message better to understand and giving error instead of SIGSEGV
2018-08-30 09:42:45 +02:00
680c9e11d4
segmentation fault in cause of empty list
2018-08-29 08:26:28 +02:00
7ecb7689f1
Intel compiler failed with SIGSEV
...
derived types, pointers, finalize .... altogether seems to bring both Compilers to their limits.
I cannot see what was wrong before, but now it works and might be a little faster
2018-08-23 00:13:57 +02:00
8c5f3d4e07
only needed once
2018-08-22 22:52:12 +02:00
ab45818d51
seems to work now
...
anyway, nicer code
2018-08-22 14:30:51 +02:00
52088691d1
improved linked list and fixed solution for strange bug
...
Bug: Using automated LHS re-allocation for a string array that with global scope seems to cause trouble
Hence, "parse_file" works with a local string and assings only once to it
Linked_List: Now storing data in the list head also and last element is always empty.
Finalize allows simple handling of deallocation
2018-08-22 11:51:23 +02:00
8de321382f
using final seems to be the better approach
...
http://www.training.prace-ri.eu/uploads/tx_pracetmo/AdvFTN_handout.pdf
still needs in-depth analysis, even though I cannot reproduce 'terminally ill'
2018-08-21 08:14:59 +02:00
d146417abe
hot fix for terminally ill with gfortran 7.3
...
might be a bug in the compiler or in the linked list.
waste some memory at the moment...
check linked list carefully before enabling again and blaming gfortran
2018-08-21 07:41:10 +02:00
9360c171a1
polished for merge
...
access to array(n+m:) is safe for array of size n with m>1
2018-08-04 19:39:50 +02:00
38183fdbf6
Merge branch 'development' into 36-faster-file-handling-for-material-config-use-stream-access-instead-of-line-wise-reading
2018-08-03 08:57:44 +02:00
bc3f6ae97c
missing bits in getStrings function
2018-07-31 23:15:44 +02:00
fb1265db3d
checking for existing file
2018-07-16 20:08:31 +02:00
a1ad18c88a
256 characters is enough for material.config
...
larger values waste memory and decrease readability.
Still need to discuss how geom files are handled, for them longer limits make sense
2018-07-16 12:08:22 +02:00
b26c4a39ef
store raw material.config
2018-07-16 11:42:58 +02:00
2bee95bfee
argument for checking expected shape in the future
2018-06-27 18:04:06 +02:00
78f9a107fe
exception for empty list
2018-06-27 09:04:14 +02:00
fc83a76318
consistent naming
2018-06-26 20:54:54 +02:00
ea80e04ea8
deallocating config, simplifying
2018-06-26 20:50:06 +02:00
1abe747204
possibility to deallocate material.config parts
2018-06-26 20:30:41 +02:00
a68c466fc6
simplified
2018-06-26 19:18:23 +02:00
6a75021bca
need to check at the beginning
...
wrong logic in case default was overwritten
2018-06-26 19:09:08 +02:00
92bcf3a7aa
function to free elements of the string list.
...
Note: Pointers that are allocated will never go out of scope. Pointers
that are assigned (=>) will be deallocated/disaccociated like allocatables
that go out of scope
2018-06-26 18:48:29 +02:00
b907acfbfa
improved if then else logic here and there
2018-06-26 11:36:45 -04:00
0680b1706f
improved error handling
2018-06-26 07:56:20 +02:00
a91fa75a28
sanity check, something is going wrong with intel 16.0
2018-06-26 07:38:41 +02:00
300b2827b2
Using new functions for parsing material config
2018-06-25 15:28:15 +02:00
c5b00e7a6c
consistent description
2018-06-21 23:48:34 +02:00
adefbd95e6
don't use intrinsic names of keywors as function names
2018-06-21 22:34:32 +02:00
b58489c1c2
merged all config related data into the config module
2018-06-19 23:53:14 +02:00
9c12ce5539
more general name (should include parsing of debug and numerics)
2018-06-14 06:39:49 +02:00
Martin Diehl