Positions on the previous file record according to a search item. The record is not read
The number of the current record is modified when it is returned by HRecNum
. For the functions that handle the current record number (HDelete
, ...), the value of this number is not updated: you must use HRecNum
(). For example: not to do:
The reading is performed from the greatest value to the lowest value of search item (see remarks
for more details).
Caution: The record loaded in memory is not modified. The HFSQL variables (Customer.Name for example, which means the Name item of Customer file) are not updated.
In most cases, HPrevious is used to position in the data file during a browse loop.
Several cases may occur after the call to HPrevious:
- the data file is empty or no record corresponds to the filter (defined by HFilter): HOut returns True.
- the function tries to lock a record that is already locked in read-only: HErrorLock returns True and HOut returns True.
The management of locks is performed on the HFSQL Client/Server data files and on the data files handled by Native Access. A HFSQL Mobile record cannot be locked. Indeed, the operating system of Pocket PC does not allow you to lock records.
Access by JDBC: The management of locks is not available for the databases accessed by JDBC.
Note: From version 19, HFSQL is the new name of HyperFileSQL.
WHILE HOut() = False
// Process the record
<Result> = HPrevious([<File Name> [, <Name of Search Key Item>] [, <Options>]])
- True if the positioning was performed,
- False if an error occurred. This problem can be caused by:
- a positioning problem (empty data file, ...): HFound returns False and HError returns 0.
- an error: HError returns an integer other than 0. HErrorInfo returns more details.
<File Name>: Optional character string (with or without quotes)
Name of HFSQL data file. If this parameter corresponds to an empty string, HPrevious handles the last data file used by the last function for HFSQL management (starting with the letter H).
<Name of Search Item>: Optional character string (with or without quotes)
Name of key item used to browse the data file. If this name is not specified, HPrevious handles the last search item used on this file by the last function for HFSQL management (starting with the letter H). If this item does not exist, the best search item is automatically used.
<Options>: Optional constant
Configures the lock and the management of duplicates performed on the record selected by HPrevious:
|hLockWrite||Lock in write mode: the selected record can be read by another application but it cannot be modified by another application.|
|hLockReadWrite||Lock in read/write: the selected record cannot be read or modified by another application.|
|hLockNo||No lock (even if HStartLock was called): the record can be read or modified by another application.|
|hDistinct||If duplicates are found, it is used to position on a single record among the duplicates. This parameter is taken into account only if the browse is performed on a key item. |
By default, all the duplicates are browsed.
The management of locks is performed on the HFSQL Client/Server data files and on the data files handled by native access. A HFSQL Mobile record cannot be locked. Indeed, the operating system of Pocket PC does not allow you to lock records.
Access by JDBC: This parameter is ignored.
The lock options will have no effect if the locks are not supported by the OLE DB provider or by Native Access.
The lock options specified by HPrevious
will be ignored. The lock mode specified by HFirst
will remain effective during the calls to HPrevious
To modify the lock mode, you must use:
For Native Oracle Access, a different lock mode can be specified for each record. However, if a transaction was started by SQLTransaction
before implementing the lock, the lock will be released at the end of transaction only (SQLTransaction
associated with the SQLCommit
The lock options are ignored. Use the lock functions (HLockRecNum
) kept for backward compatibility.
Read operation according to a key
HPrevious positions on the record with the greatest key value.
The sort order is the one that was specified in the analysis for this key.
Comparing HPrevious and HReadPrevious
does not read the record: therefore, HPrevious
is faster than HReadPrevious