Problems resolved by CA-Clipper 5.2c:
1. Fixed most instances of the DBFNTX/1210 error.
2. Fixed INDEXKEY() so that when it is called with an invalid order argument it will no longer corrupt the data at memory location 0000:0000 in dos and will no longer generate a General Protection Error in Exospace.
3. Fixed CMEM.OBJ (in Clipper.lib) so that the malloc(), _fmalloc(), free() and _ffree() memory allocation functions will return the correct information.
4. Fixed VM Integrity error when evaluating detached code blocks that are nested three or more deep.
5. Fixed VAL() so it no longer left justifys the result.
6. Fixed ACHOICE() so that it will not redraw the menu window after returning from the user specified function unless: A. the user specified function changes the number of elements in the menu array. B. the user specified function returned the new return code of AC_REDRAW (which can be found in ACHOICE.CH and has the numeric value of 4).
7. Fixed the Expanded Memory Manager so it will not attempt to use more than eight megabytes of EMM. Currently, Clipper can not take advantage of more than eight megabytes of EMM. Please note that the printed documentation is incorrect. This was the cause of many corruptions on systems that had more than eight megabytes of expanded memory.
8. Fixed an incorrect calculation in _xvalloc() that caused the Virtual Memory Manager to allocate 1K to much if the requested size + 16 (in bytes) was an even multiple of 1024.
9. Fixed an internal calculation that did not account for overflow when converting a segment:offset address to an absolute address. This could only be encountered by calling the VM API function:_xvalloc().
10. Fixed several problems in the DBFCDX replaceable database driver.
11. Fixed a compatability problem between CA-Clipper Tools-II and CA-Clipper 5.2x. Developers using CA-Clipper Tools-II noticed unresolved symbols at link time. The CT2PATCH.OBJ file will resolve these problems and should be included as an object on the link line. Please note that CT2PATCH.OBJ is only intended for use with CA-Clipper Tools-II, those not using CA-Clipper Tools-II should not include this file in their applications.
12. Fixed DBU so that closing a file with an associated filter will no longer cause a DBCMD/2001 error.
13. Fixed DBU so that performing a replace will no longer cause a "Lock Required" error.
14. Fixed REPORT FORM so header will print when creating a report whose width is greater than 254 characters.
15. Fixed REPORT FORM so that it now ejects properly for groups.
16. Fixed REPORT FORM so that double spaced forms will now print properly.
17. Determined that the EG_SYNTAX error in REPORT FORM was not caused by an empty database. It was caused by the absence of a field name for the report. This is the correct behavior.
18. Fixed REPORT FORM so that it will respect the "SUMMARY ONLY" option when requested.
19. Fixed REPORT FORM so that it no longer prints an extra form feed after the report is completed.
20. Fixed RL so that it will save entered data from all entry screens instead of only the current display screen.
21. Fixed RL so that it will no longer save the rightmost column when it is empty.
22. Fixed RL so that it accepts "T" or "F" in it's question fields. It previously was limmited to "Y" or "N".
Technical Note: RLOCK()/DBRLOCK() problem
o Problem: RLOCK() and DBRLOCK() (without any parameter) will not reliably lock a record under certain circumstances, even though their return values (.T.) indicate success. This occurs when locking a record in the child database of a relation, after the record pointer has been repositioned by a SEEK or GOTO, but before the data has been accessed.
o Solution: A workaround for this problem is to use DBRLOCK( RECNO() ) instead of RLOCK() or DBRLOCK(). A simple way to implement this is:
1. Add the following statements to a copy of the std.ch header file: #translate RLOCK() => DBRLOCK( RECNO() ) #translate DBRLOCK() => DBRLOCK( RECNO() )
2. Re-compile your program with the /u
o EXAMPLE: use child shared new set index to childntx use parent shared new set relation to FieldOne into child goto 2 ---------------> if ( rlock() ) // change this rlock() to dbrlock(recno()) replace child->FieldTwo with parent->FieldTwo endif close data return
o Note: This is a temporary workaround while a solution for the problem is being worked upon.