Minor updates to the documentation. HTML documentation index now
in
<prefix>/share/doc/MEMCOM/index.html
.
Fixed an error in the Array Tables when initializing data.
Updated documentation: MemCom is not thread-safe.
Fixed a problem with mcTBstore
: Table size
was updated in some cases.
Make the mcSTload
function public for
pymemcom
Errors in documentation of mcf_tb_extract
corrected.
Changed RTABLE_DESCR_LENGTH
from 1024 to 64
and RTABLE_DESCR_LENGTH_INCR
to 64. This makes
database files smaller.
Added the function
mcDBsaveFileNoSync
Function mcf_tb_rep
renamed to
mcf_tb_replace
(as in documentation.
mcf_tb_rep
still available for
compatibility.
mcTBloadAll
: fixed performance problem for
ST datasets.
Documentation updates: Improved print functions: dataset names
now printed in full length. Corrected a documentation error in doc of
mcf_tb_put
. Corrected documentation error in
mcf_db_delete_set
.
C interface: Function mcDBopenFile
now
returns correct error status from 'change working directory'
function.
FTN interface: Subroutine
mcf_db_get_set_att_size
returned status 1 instead
of 0 in case of correct termination. Fixed.
Sparse tables: API and implementation change. Documentation published. Improved the table pack algorithm.
C API: DMM function mcDMMdel
: The manual
says that the user code ident=0 is legal but the C code did not test
properly. Fixed.
F77 interface: A problem appearing occasionally with the
relational table size in subroutine clrtb
has
been fixed.
Tools: The utility program mcchecklicense
is now documented.
FTN: Fixed some problems with return status checks in test programs.
The sparse table (ST) API has been added.
An problem with the old Fortran f77 API (library libmemcomf) and clearing relational tables has been fixed: In some cases the call to clrtb did not restore the correct default table size of 8192 bytes. Note: The Fortran ftn API (libmemcomftn) is not concerned by this problem, because the table access mechnism is different.
An problem with the old Fortran f77
API (library libmemcomf) with the subroutine
dmmdel
has been fixed: The user code set to 0
now works as documented.