This table lists the most recent releases of Texpress 8.0 for each platform.
Select the Number in the Version column for Release Notes.

Top

Release Notes

8.0.047 - 22 June 2009

• Fixed issue where query timeouts did not function correctly for searches that sorted the results.

8.0.046 - 03 October 2008

•  Fixed problem with generation of temporary file names.

8.0.045 - 04 December 2007

•  Fixed problem where a deadlock error may occur when using texload and texserver/texql to update records in the same table.

8.0.044 - 04 August 2006

• Altered query auditing so that it now records a texql statement of the query performed.
• Added new texpress option auditident which allows for the recording of a unique identifier against each query and the records displayed from that query.
• Altered texaudit to accept -i as a new parameter which will display identifier information for queries and record displays.

8.0.043 - 19 May 2006

• Updated column format setting for incrementation fields to ignore case based formats (i.e. uc, lc, etc).
• Fixed problem with unique value checking where an insert on one cursor with an edit on another cursor before saving the insertion may result in duplicate unique values being allowed.
• Fixed problem where clearing a nested table field and saving the record without making any other changes may result in an error indicating a value is not unique, when it is unique.

8.0.042 - 26 April 2006

• Corrected unique value assignment issue with simultaneous access by multiple users. A New record operation followed by Cancel record without saving operation could result in a unique number being skipped.

Top

8.0.041 - 30 January 2006

• Changed default display formats for latitude and longitude items to use dd mm S D. The new format allows arbitrary precisions to be displayed correctly.
• Fixed problem with dateorder() interpreter function where an order of "ymd" resulted in "mdy" being set.
• Reworked latitude/longitude code to provide consistent handling of arbitrary precisions, particularly when converting from decimal degree or decimal minute formats.
• Increased maximum viewable history records to 100 for texforms.
• Fixed problem where multiple uses of the same cursor (via a TexReset() call) may result in less than optimal index matching.
• If interpreter expressions use the system() call it is possible that the match array may become corrupted, possibly leading to a large increase in data file size.
• Updated locking code to ensure that all buffers in locked file are flushed after the lock succeeds and before the file is unlocked.

8.0.040 - 28 December 2005

• Added COMPUTED flag to fields that contain values that are computed (via assignment or validation expressions or via a link database). The computed flag is dumped in the XML audit trail so that standalone loading does not try and load the field.
• Fixed problem with newmonths() interpreter call where passing an empty day value causes the previous month to the be returned rather than the correct month.
• Fixed problem where clearing all Partial Index values in texdesign would incorrectly turn off all item flags except for IDXPART. The IDXPART flag is now reset correctly and other flags left on.
• The standalone import code has been modified to correctly translate Unicode based XML into Latin-1.
• Fixed problem with destroying a texql object in texql.pm before issuing any command. The server now terminates correctly.
• The standalone import routines now ignore all computed fields when loading data.

8.0.039 - 23 November 2005

• Adjusted texload to obey uniquestrict option settings and to implicitly set uniquestrict=false when the texload -k option is used.

Top

8.0.038 - 14 November 2005

• Support for Solaris 10 on AMD Opteron x86_64 added using Sun Studio compiler.
• Cleaned up qsort() prototypes to cut down on compiler warnings.
• Changed handling ofMDY data format so that dates like Jan 2005 are interpreted as month/year rather than month/day.
• Fixed problem with 64 bit forms that contain partial indexing attributes.
• Changed texdesign so that non-existent link tables that have internal ids already allocated generate a warning rather than an error. The form can still be saved correctly (using the already assigned internal ids).
• Adjusted query count supplied by server to update correctly for queries that are not exact matched.
• A key query with multiple keys will now return the matches in the order specified in the query statement.
• Added new command, Refresh Record, to texforms. The command allows the latest version of a match to be displayed.

8.0.037 - 22 August 2005

• If mempool allocation fails, decrement memory size by 1Mb at a time rather than 1 Kb.
• Added new command, texreadonly. The command sets write locks on a supplied table so that the table is put into a read-only state for the length of the command (works similar to texlock).
• Modified stack dumping code to work on FreeBSD.
• Modified "Value as Text" export setting to function correctly for double nested grids.

8.0.036 - 5 July 2005

• Added Also Search code back after accidentally removed in 8.0.031.
• Reworked latitude/longitude calculation code to ensure correct precision is maintained.
• Fixed rounding error in centroid calculations.
• Large improvement in form validation time added.

Top

8.0.035 - 23 June 2005

• Removed texsdiagent from Texpress source. Support for search profiling is no longer available.
• Rewrote code that atomises search terms so that the same code is now used by both texforms and texserver. The strings used to break words and numbers up have been removed (TWORDBRK and TNUMBRK) allowing the texserver TWORDJOIN string to define what delimits a word.
• Support for partial word indexing added. Indexing is now available for patterns that have leading letters.
• Index support for NULL and NO NULL added.
• Support for UWIN platform removed.
• Fixed handling of literal next character "\" for texql LIKE based queries.
• Unique format strings extended to allow "%s" format to specify acceptable punctuation characters. The extended format is "%[punctuation]s".
• Improvements to speed of range queries made.
• An XML import facility has been added to texserver.
• Fixed problem with memory leak in PAM authentication. Also added support for multiple simultaneous authentications from different PAM sources.
• Fixed possible memory corruption issue with internal timer (only triggered if query upcalls used).
• Fixed problem with use of "\n" between rows in a table, "\r\n" is now used consistently. The change allows embedded "\n" in data to be interpreted correctly.
• Fixed problem in texql optimiser where complex statements may not be parsed correctly.

8.0.034 - 21 April 2005

• Tidied up odd strings sent from standard validation expression on the "Date" library item.
• Changed century calculation so that it is only done once.
• Fixed incorrect field length calculation for texexport() call for fake nested tables (id0, id1, etc).
• Fixed problem with centroid code where bad data values are supplied for centroid values.

8.0.033 - 5 April 2005

• New expression functions added, centroidstart, centroidadd, centroidset to compute centroid values.
• Fixed rounding problem with Latitude/Longitude when using the dd mm.mm format.
• Fixed problem with Latitude/Longitude and Time values where precision was sometimes lost.
• Efficiency mechanism put in place to streamline handling of empty date columns.
• Unique incrementation format string enhanced to allow an optional non-incrementing component.

8.0.032 - 23 March 2005

• XML auditing adjusted to correct occasional logging error when new records were inserted.

Top

8.0.031 - 27 January 2005

• Altered texconf to accept up to 8 digits for the number of segments.
• Corrected analyse index terms counting code.
• Added entity translation to the XML auditing.
• Corrected row update problem where a race condition under certain stress conditions allowed two successful edits on the same record.

8.0.030 - 29 November 2004

• Fixed XML auditing problem where an empty xmlauditemslist caused no auditing information to be captured.

8.0.029 - 20 November 2004

• The correct encoding type is now specified in XML documents produced.
• Fixed XML audit log problem experienced when runningMicrosoft Services for Unix.
• Fixed problem with internal latitude/longitude encoding where in some circumstances a rounding error occurred when value precision was not set. This problem made range queries ineffectual on latitude/longitude values where the operator was .
• Fixed buffer error in the valdate1, valdate2 and valdate3 expression functions. Under some circumstances the functions returned unexpected results. These historical date functions were superseded in version 6.0 by the DateValid function.
• Fixed problem with texdfdiag program not picking up unusual data history record offsets.
• Fixed obscure unique file locking problem that only arises under a certain sequence of events. Caused by incorrect code trying to move a unique value lock set by another process. Unique value mechanism now accommodates the possibility of occasional internal dummy values in the unique file.
• Fixed texdesign schema problem with the copying of single field library items.
• Adjusted timer functionality introduced in 8.0.028 to address certian situations where the timer may trigger too early.
• Perl wrapper for Texql query language, texql.pm, adjusted to a) provide more robust parsing of integer values; b) rectify problem with returns arrays of values.
• Fixed export problem with setting of "alias" name for a tuple column.
• Improved robustness of query timeout facility introduced in 8.0.028.

Top

8.0.028 - 7 September 2004

• Added query timeout facility. New application option timeout=x where x is the number of seconds a query may run before timing out. The timeouthandler option can be set to one of error, return, continue or ask. Where:
  • error - generate an error message
  • return - return matching records so far without an error
  • continue - ignore the timeout and continue processing
  • ask - perform an upcall and ask the user what to do
• Added searchlimit=x database option that can be used to indicate the maximum number of segments that may be part of a search. the number can be an absolute number of segments or a percentage (eg. 20%). The searchlimithandler option can be used to indicate how the limit should be handled. It takes the same values as timeouthandler.
• Added searchupdate database option that provides feedback to a user about the number of records retrieved while searching. The update period is number of seconds between each update call.
• Added TexOptionSet() API call that can be used to set database and application options via a user program.
• Added TexCursSearch() API call to allow upcalls to be performed while a search is underway.
• Added ! operator to xmlaudititems specifying that an item should always be audited.
• Rewrote the SIGALRM handling code to allow multiple timers to be active at the same time.

8.0.027 - 18 August 2004

• Updated character map ISO-8859-1 (Western European) to contain euro symbol.
• Updated character map ISO-8859-7 (Greek) mappings.
• Added new character map for ISO-8859-5 (Cyrillic), ISO-8859-6 (Arabic), ISO-8859-8 (Hebrew), ISO-8859-9 (Latin 1 with Turkish support), ISO-8859-11 (Thai).
• Addressed issue where a forced save of a non-modified record encountered problems if there were unique value constraints.
• Additional large logical file debug added for various platforms.

8.0.026 - 12 August 2004

• Added new character map for ISO-8859-7 (Greek).
• Additional TexAPI functions added for Standalone support.
• Corrected problem when "readonly=yes" option was set at the application level.
• Adjusted database diagnastic tool, texdfdiag, to provide output that distinguishes between saved History records and records that have been modified since last exclusive maintenance.

Top

8.0.025 - 6 May 2004

• Added new API call TexCombine(int cursor1, int cursor2, flag) that can be used to combine the results of cursor2 into cursor1. The combination flag can be a merge, intersection or difference.
• Fixed problem with long path names (> 127 characters) where server may crash.
• Corrected minor problems with texparams and texupgrade introduced in 8.0.023.

8.0.024 - 6 May 2004

• Fixed problem where audit generated through texserver would not be correct if different databases have different audit options set.

8.0.023 - 26 March 2004

• A new database option indexpath is available to set the directory path for all database index files. This makes it easier to keep database and index files in separate locations. Specifically the index files/directories are dist, key, lut, oflow, rec, seg, unique.
  - indexpath can be set in all the usual places, TEXHOME/etc/opts, data/dbname/opts, TEXPRESSOPTS, dbnameopts; the order listed here is the order of evaluation.
• The filepath database option is no longer supported. Use dbpath instead.
• A new expression interpreter function sort is available that will sort a series of columns as though they were one nested column. The arguments are a list of column names that make up the grid, with the sort columns expressed as strings with a leading + or - to indicate the sort required.
• Enhancements to unique incrementation value facilities to handle alphanumeric components and provide more flexible parsing to accommodate more intricate format specifications. Also special coding added to permit incrementation triggering of a single unique column of integer type by assigning the value -1.
• Fixed problem with old Texpress 5.0 date/time library items. In Texpress 5.0 automatic values had zero padded two digit month/day and hour/minute. In Texpress 8.0 the values were not zero padded. Adjusted to have same behaviour as Texpress 5.0.
• Fixed problem with API call TexColFetch().
• Fixed problem where in certain situations texql allowed assignment of a primary key value without first checking that the key value was available.
• Fixed problem where if a row insert failed because the database configuration was full (and no dbname/oflow was in use) the row data would still be written to the data file.
• Fixed problem with unique value assignment where under certain circumstances cancellation of a record edit could allow an immediate subsequent record insert to re-use an existing unique value.

Top

8.0.022 - 22 January 2004

• Fixed problem where exact match count was incorrect because count was not adjusted when duplicate matches were removed.
• Added ability to audit in XML. This auditing encompasses all previous auditing options and gives additional options to audit changes to a record. See How do I get XML auditing of database actions for more details. 
• Fixed problem where characters 128-159 in the data file could cause an infinite loop.
• Altered texql.pm to correctly handle an undefined value.
• Fixed problem with texql where extended characters would not print correctly when a dumb terminal type was used.
• Fixed problem in texserver where long words (>128 chars) would terminate when exact matching.
• Fixed bug in summary sorting where a sort of greater than 3 levels produced incorrect results.

8.0.021 - 10 December 2003

• Parsing and storage of precision for latitudes and longitudes altered to function correctly.
• Fixed a file descriptor leak (/dev/tty) when a process was exec'd. 
• Altered texql.pm to accept different float formats.
• Fixed report summary to use numeric sort when numeric fields are used.

8.0.020 - 9 December 2003

• New option nchsdates added that allows recording of special 00 and 99 values in dates eg. 00/12/2003 or 99/99/9999.
• New option dateswitch added which allows a point to set (mod 100) at which a two digit date years rolls to the next century.
• Fix in file access routine to correctly return file permissions.
• Fixed column data validation to ensure that only valid file characters were allowed into text items, and only valid numeric characters were allowed into integer and float items.

8.0.019 - 26 August 2003

• texaccess modified to allow user privileges to be set from the command line.
• Problem with error messages containing duplicate error strings resolved.
• If the option uniquestrict is set, but a format has not been set via uniqueformat then texenv -v dbname would dump core. The problem has now been fixed.
• Fixed a file offset overflow problem in texpatch.
• Fixed a problem where texforms would dump core if a table had the number of words per segment descriptor set to zero.
• Fixed incorrect highlighting of matching terms when a texql ORDER or UNION statement was used.
• Corrected problems where a texql statement containing a key query ORed/ANDed with other conditions would evaluate incorrectly.
• texql.pm now has support for boolean types.
• A possible buffer overflow with long "exec" commands has been fixed.
• Fixed problem where texserver returns unpredictable results to TexRowFetch() where there are no matching records.

Top

8.0.018 - 24 June 2003

• Modification to Unique value commit to allow texserver to be able to save a non pending Unique value. Required in certain situations when Unique value is set via back-end form validation.

8.0.017 - 20 June 2003

• Potential race condition under heavy load addressed in Primary Key value validation.

8.0.016 - 19 June 2003

• Internal reworking of TexAPI function TexRowRefresh() to provide additional error reporting.
• Added additional error reporting to better track the duplicate Unique value checking process.
• Potential race condition under heavy load addressed in Unique value validation.
• Fixed internal Primary Key value and Unique value issue which under heavy load may result in a key state being incorrectly read.

8.0.015 - 1 June 2003

• New column summary server-side facility provided for use with TexAPI.
• Fixed locking problem in texload with key allocation when -s option used.
• Fixed "licence limit exceeded" error logging issue in texserver.
• Fixed problem in string printing code where %*s construct was used.
• Fixed problem in texload making automatic Unique value triggering of a single integer column difficult.
• Database option loadmemory default changed to 16Mb.
• Database option autokeytries default changed to 1,000,000.

8.0.014 - 15 May 2003

• Changed uniqueformat syntax for multi-column unique values to use %& instead of %| as the column separator. Avoids export/import issue if the EMu emuregdump/emuregload or Vitalware vwregdump/vwregload scripts are used. Any uniqueformat back-end settings or client side column Format settings in EMu/Vitalware that use %| must be changed to use %&.
• Adjusted unique value processing for multi column unique values to ensure correct EMu/Vitalware client side display occurs when a client side format is used similar to:
  Group|Default|Table|table|Format|columnname;columnname2;..|format
• Improved error handling for bad uniqueformat strings.
• Fixed problem where an upgraded database user file contained user names from other hosts that were not valid user accounts on the new host. These usernames were incorrectly denying the addition of a * (star) user account for the database.

Top

8.0.013 - 9 May 2003

• Added support for PAM user authentication.
  Typical /etc/pam.d/texpress file for Linux is:
  auth required /lib/security/pam_nologin.so
  auth required /lib/security/pam_stack.so service=system-auth
  account required /lib/security/pam_stack.so service=system-auth
  session required /lib/security/pam_stack.so service=system-auth
  password required /lib/security/pam_stack.so service=system-auth

• Rework of unique value insert/update/delete code to rectify various problems and fix flawed strategy for texserver process running multiple cursors on same module.
• The texufdump diagnostic program updated so that the -z option will provides additional diagnostic information from the unique value files.
• Data file garbage space re-use code re-worked to ensure the dbname/free file is portable across all Texpress platforms. Portability problem arose due to internal structure size differences on Solaris SPARC and 64-bit platforms.
  NOTE: On any Solaris SPARC or 64 bit platforms such as Digital Unix when you upgrade from pre 8.0.013 to 8.0.013+ the dbname/free file for every database must be cleared. Non Solaris SPARC 32 bit platforms do not need to have the dbname/free file cleared but there is no harm in doing so if you are unsure whether this step applies to your situation.
  Whilst no Texpress processes are running do:
  cd data
  for i in */free
  do
   echo $i
   cp /dev/null $i
  done
• If you are upgrading from an earlier version of 8.0 (i.e 8.0.n where n < 13) then prior to first user access post upgrade you must rebuild the index for all databases that have one or more Unique value column(s).

8.0.012 - 1 April 2003

• Certain texql query situations involving many OR clauses each with many AND terms were resulting in high numbers of index false matches. Several indexing adjustments made to address this issue.
• Added -d option to texql program to provide better status information for insert, update and delete data manipulation (DML) statements. When this option is used a single status row of three atomic values is output in the format: (OkayRows, FailRows, FailMessage)
  OkayRows and FailRows are integer, FailMessage is text. If FailRows is non-zero then FailMessage will state the first error encountered during the DML command.
• The texql.pm script has been modified to provide access to DML command status information (described above) via its standard Row and Col calls.
• Fixed problem in texql optimiser for obscure query that degenerated to FALSE.

Top

8.0.011 - 8 March 2003

• Fixed problem that caused texupdindex to ignore the loadmemory option setting and unbitslice/rebitslice the index using a very small amount of memory. Problem only occurred where the database logical seg file size was > 2Gb.
• Enhanced texql to recognise Crystal Enterprise SQL date format: { d 'yyyy-mm-dd' }
• The TexRowFetch() call would not correctly return data for a double nested grid where the lower level was not merged into the upper level (that is a \n between rows). The correct data format is now generated.
• The TexRowFetch() call allows a format to be specified on the end of a column name (eg. DatDate:dd/mm/yyyy).
  The format is used when fetching data for that column. The code did not correctly handle the case where a ":" character occurred in the format string (eg. TimTime:hh:mm). The format is now handled correctly.

8.0.009 - 21 February 2003

• New Append Fields command added to texdesign to allow multiple fields to be added to an item in one operation.
• Fixed problem in texql.pm where multiple invocations left zombie processes.
• Fixed problem where some texforms Temporary Record saves later resulted in the error message: "Key value was taken by another user (beaten)" being displayed when the temporary record was attempted to be retrieved. Also the texforms "List temp records" Insert mode function just listed (-1) for these records. Caused by initialisation problem when the first operation performed in a texforms session is a Temporary Record save.
• Fixed problem with some old Texpress 5.0 date/time library items. In 5.0 automatic values had zero padded two digit month/day/hour/minute. In 8.0 the values were not zero padded. Now adjusted to have same behaviour as 5.0.

8.0.008 - 10 January 2003

• Fixed problem with unique value incrementation occasionally caused when a record edit is cancelled.
• Default database configurations now take a more optimistic approach resulting in smaller initial index sizes.
• Fixed error with use of getditto expression function on a Query form.
• Fixed up problem in texql CONTAINS clause where a closing parenthesis is in the query terms.
• Fixed problem where texql insert statements were not assigning any form defaults values.
• Fixed a problem with texql words() function.
• Fixed a problem in texupgrade when upgrading a Texpress 5.0 database to 8.0.
• Problem with monthl expression interpreter call fixed.
• In texql.pm added support for negative integers and floats.
• New database option added called uniquestrict. Can be set to either true (default) or false. A setting such as uniquestrict="CatRegNo=false" will flag all non unique values but permit them to be stored.

Top

8.0.007 - 4 October 2002

• Additions to unique value auto incrementation facilities.

8.0.006 - 3 September 2002

• dateswitch=nn
  option added to provide more flexibility in parsing of legacy 2 digit date values. If the two digit year value is greater than the dateswitch value then the date is taken as being in the previous century.
• Problem with automatic allocation of incrementing key values rectified.

8.0.005 - 28 August 2002

• Support for texvideo removed from Texpress.
• A new attribute allows an item to be defined as unique. When a record is saved and a unique value is not found in the item, validation fails.
• The ability to have Texpress automatically allocate incrementing key values has been added. A format string can be specified on an item (which must also have unique turned on) that defines the part(s) of the key that are incremented.
• Default date format changed to dd/mm/Y e so that BC is printed for all dates before year 0 AD.
• Support forMicrosoft Services for Unix (SFU) added to Texpress code.
• texclose and texopen modified to allow multiple tables to be specified on the command line.
• Fixed Bad number of words in record descriptor problem with 64 bit configurations.
• Fixed problem with validation of BC dates in interpreter.
• Fixed bug with large sorts (over 800,000 records) in texserver.
• Fixed up a problem where Texpress could fall over when reading a corrupted record.
• Fixed bug when COPYALL and SAVEALL commands are used in texforms, where incorrect records could be deleted.
• Support for new utmpx structure added to Texpress.
• Fixed up problem where tty structure many not be reset correctly in Texpress.
• Fixed bug when reading double nested grids where all the nested rows are empty.
• Changed texserver so that TexRowHits() returns the number of hits on the underlying table, even if the query only projects a limited number of records.
• Changed texserver exact matching code so that ^term and term$ are checked at query time, rather than as the records are shown. Removes displaying of bad TexRowHits() value.
• Fixed bug where texserver may fall over when updating statistics on library items.
• Fixed bug where irn = 1 and irn = 1 would result in two matches.

8.0.003 - 27 March 2002

• NOTE: Upgrading to KE Texpress 8.0 will require the texupgrade program to be run on each database (as the DBA). KE Texpress 8.0+ uses new formats for many database files. These file formats cannot be understood by pre KE Texpress 8.0 programs.
• Support for individual database files > 2Gb in size has been added to KE Texpress 8.0. This support applies to all files used by KE Texpress. The implementation uses numerous physical files to implement one large logical file. Each file partition is a maximum of 2,147,418,112 bytes in size. The first partition is the base file name (eg. data) while successive partitions have a # followed by the partition number appended to the base name, zero padded to 3 digits (eg. data#001, data#002, etc.). There is no limit to the number of partitions that can be created. All KE Texpress code has been modified to use 64 bit file offsets for data offsets stored in the data, key, rec, temp, lut*, uniq*, free, oflow files. The new maximum logical file size is now 2^63-1 (or 9,223,372,036,854,775,807) bytes.
• Note this large file support is available on all 32 bit platforms as well as 64 bit platforms. A KE Texpress 8.0 database with large files can be copied to any other KE Texpress 8.0 platform and run without modification.

Top