FLAM® Issue Tracker

FL5 - Change Log

FL5 - 5.1.24 (Released 2021-02-01) View Issues ]
==================================


5.1.24-29930

 * Ensure dynamic link of libssh after change from SCLM to make file build on z/OS

5.1.24-29928

 * Bugfixes
 ** Prevent segmentation fault (0C4) of subtask (FLZSRV) in FLUC subsystem if vdFcrPut failed (if open of file not possible) in vdFcrFlu
 ** Prevent wrong instruction (0C1) in PCR routine of zIIP support based on missing resources if connect to WLM failed
 *** Use ASCBJBNS/NI as WLM Subsystem Name

5.1.24-29911

ATTENTION: Because of huge memory consumption the defaults object on
column level was removed from table support. Please define the
corresponding values directly for these columns.

 * Support tag value delimiter format in table support
 ** The new general text format parser was built mainly to support SWIFT MT formats
 ** It can also be used to parse any kind of hierarchical text format including XML, JSON and others
 ** It can be used for a set of columns inside other column formats (partial parts with complex structure)
 ** This allows to parse fixed or variable length formats in between a piece of JSON or XML
 * CPACF cryptographic acceleration for the 64 bit environment on Z systems is now supported
 * Improved system information for hardware acceleration with CPACF on z/OS and USS
 * Support ignoring a prefix in front of colon for comparison of XML tags in table support
 ** The same row specifications can be used if different name space abbreviations used
 ** The new parameter PREFIX in FORMAT.XML() (or XMLPFX in the default definitions) can be used
 * Support "ROOT" as alias for "PATH" in table and row level
 * Reduce memory consumption for header values at XML parsing in table support
 ** No growing of memory through collection of header values
 ** The memory usage is now constant independent of the amount of data converted
 * Improve and correct reading of binary delimiter
 ** The record buffer is no longer used to store the delimiter (the full record length are used for record data)
 ** Implement special function for 1 and 2 byte delimiters to increase performance
 * Add more PL1 samples (SPFCUCNV using FLUCUP, SPFLCICV using FLCICV (iconv like DLL of FLAM))
 * Add FLAM4TDA as part of FLM4SEPA to the package
 ** Check feature code in assembler modules
 * Initialize dia-critical characters support also if quiet is defined for the command line executor
 * Improve performance in whitespace handling of ICV character conversion module
 * Support enforcement of 32 or 64 bit ZIP archive format (prevent 7z warning)
 ** ARCHIVE.ZIP( ... FORMAT=Z32/Z64 ...)
 ** If no format is given the old behavior is used
 ** The default streamed format now is Z64
 * Implement ICV character type check functions isICV - upper/lower/punct/space/cntrl/digit/xdigit
 * Support 64 bit CCA support program with 64 bit integers on platforms where the C data type long has 64 bit
 ** The requirement to use 64 bit instead of 32 bit integers is determined by testing of the library
 ** This ensures that the SAPI library is handled correctly
 * Reduce minimal keyword length for new FLAM4 executable to be more backward compatible with the old FLAM utility on Windows
 ** 100% backward compatibility is not possible, because of newly supported keywords
 * Make implementation of new license modules compatible to old FLAM versions
 ** Using a new license module with an older FLAM revision could result in a segmentation fault
 * Support enforcement of post-processing per file
 ** By default the post-processing of the I/O components was done only once after closing all files
 ** This resulted in only one post-processing call for the last file during directory walk
 ** With the new switch 'PERFILE' post-processing per file can requested
 * Correct reset of environment variables (FL_PLATFORM) for Windows
 ** Use UNSETENV instead of SETENV with empty string if it was no variable defined before
 * Correct mapping on Unix and Windows platforms in new FLAM4.exe and FLCL FLAM command if no record format is given and a record delimiter is specified
 ** Since 5.1.23 the record delimiter was only relevant if INRECFORMAT=TEXT/STREAM was specified
 ** Since 5.1.24 if no record format but a record delimiter was specified, then the record format will set to STREAM by default
 * A new section in the FLAM4 manual and install.txt describes how to specify input and output record format as property
 ** Unify behavior of the new flam4.exe with the old one on Windows and Unix platforms
 * Add symbol table and memory allocation statistics for CLP based on the environment variables below
 ** CLP_SYMTAB_STATISTICS - Print symbol table statistics to STDERR stream when closing
 ** CLP_MALLOC_STATISTICS - Print memory allocation statistics to STDERR stream when closing
 * Add new parameter BLK2REC to WRITE.RECORD() and WRITE.FLAM4() of CONV command
 ** Add an additional delimiter based block to record conversion with suppression of trailing whitespace
 ** Required to convert for example a SWIFT-MT output format to a record per line
 * Finalize each column after table change at write
 * When using SSH configuration files with "IdentityFile" lines for public key authentication,
   all matching keys are now attempted in the authentication phase.
 ** Previously, only the first encountered key was considered by libssh, all others were ignored.
 * Make separator character in table definitions compatible between read and write
 ** when reading more than one separator character (SEPCHR) can be specified
 ** when writing only the first of them is used
 * Introduce global literal caches per character set to reduce memory consumption
 * Add log of license ID also at the end of the INFO function

 * Bugfixes
 ** Fix several reallocation issues after enforcement of reallocation for internal write buffer
 *** Normally the size of the write buffer is pre-calculated and only in rare cases a reallocation is required
 *** Our test framework enforce now all this reallocations and in few components the handling was buggy
 **** Fix possible damaged data after length error occur in table support
 ***** in very raw situations the internal buffer to write XML data in table support cloud be too small
 ***** in such case a reallocation of this buffer is done and the XML writer is called again
 ***** now we ensure, that the internal state of the XML writer is correct after an re-call with more memory
 ***** in the same rare case the offset for the element converter was not correct to write CSV format correctly
 ***** in the same rare case the input pinter for the formatting was not re-corrected after a re-allocation with pointer change
 **** Fix wrong pre-calculation of required space for re-allocation in CNVBAS (prevent possible segmentation fault)
 **** Fix segmentation fault after re-allocation with pointer change and correct amount of required space in FMTXML
 **** Fix re-allocation issue in CNVPGP (the small sizes smaller than minimum space was not checked (the enforcement goes below (not really a bug)))
 **** Verify output space before fractional length for big integer processing to ensure length error
 **** Fix maximum length check in CSV-Support at write if masking requested
 **** Correct re-allocation logic in FMTREC
 ** Fix whitespace handling "COLLAPSE" in ICV module (replacement was done with data instead of whitespace)
 ** Fix dirty delimiter handling of '0D..0DC285' in FMT.TXT()
 *** The 0x85 was not removed from the data, which normally results in conversion errors
 ** Fix UTF-32 support in integer and float converters (plausibility check if length multiple of 4 failed)
 ** An issue in integer and float converters when using UTF-32 is fixed



- 0001017: [1.1 FLCL] ":32B:GBP31500," are a valid swift TAG with 31500,00 as amount (Tilo Eckert) - resolved.
- 0001019: [6. Access Control, Logging and Errorhandling] Warning for soon to expire license is never shown to user (Falk Reichbott) - resolved.
- 0000930: [1.2 FLCC] Support tag value delimiter (TVD (required for SWIFT)) formats in table support (Falk Reichbott) - resolved.
- 0001009: [1.1 FLCL] Record length requires the delimiter length, but the delimiter is not part of the record (Falk Reichbott) - resolved.

[4 issues]


Copyright © 2000 - 2024 MantisBT Team
Powered by Mantis Bugtracker