2 1.805 1st September 2002
4 * Added support to allow DB_File to build with Berkeley DB 4.1.X
6 * Tightened up the test harness to test that calls to untie don't generate
7 the "untie attempted while %d inner references still exist" warning.
9 * added code to guard against calling the callbacks (compare,hash & prefix)
12 * pasing undef for the flags and/or mode when opening a database could cause
13 a "Use of uninitialized value in subroutine entry" warning. Now silenced.
15 * DBM filter code beefed up to cope with read-only $_.
19 * Perl core patch 14939 added a new warning to "splice". This broke the
20 db-recno test harness. Fixed.
22 * merged core patches 16502 & 16540.
26 * Fixed a problem with db-btree.t where it complained about an "our"
27 variable redeclaation.
29 * FETCH, STORE & DELETE don't map the flags parameter into the
30 equivalent Berkeley DB function anymore.
32 1.802 6th January 2002
34 * The message about some test failing in db-recno.t had the wrong test
37 * merged core patch 13942.
39 1.801 26th November 2001
41 * Fixed typo in Makefile.PL
43 * Added "clean" attribute to Makefile.PL
45 1.800 23rd November 2001
47 * use pport.h for perl backward compatability code.
49 * use new ExtUtils::Constant module to generate XS constants.
51 * upgrade Makefile.PL upgrade/downgrade code to toggle "our" with
54 1.79 22nd October 2001
56 * Added a "local $SIG{__DIE__}" inside the eval that checks for
57 the presence of XSLoader s suggested by Andrew Hryckowin.
59 * merged core patch 12277.
61 * Changed NEXTKEY to not initialise the input key. It isn't used anyway.
63 1.79 22nd October 2001
65 * Fixed test harness for cygwin
69 * the test in Makefile.PL for AIX used -plthreads. Should have been
73 10372, 10335, 10372, 10534, 10549, 10643, 11051, 11194, 11432
75 * added documentation patch regarding duplicate keys from Andrew Johnson
80 * AIX is reported to need -lpthreads, so Makefile.PL now checks for
81 AIX and adds it to the link options.
83 * Minor documentation updates.
85 * Merged Core patch 9176
87 * Added a patch from Edward Avis that adds support for splice with
90 * Modified Makefile.PL to only enable the warnings pragma if using perl
93 1.76 15th January 2001
95 * Added instructions for using LD_PRELOAD to get Berkeley DB 2.x to work
96 with DB_File on Linux. Thanks to Norbert Bollow for sending details of
100 1.75 17th December 2000
102 * Fixed perl core patch 7703
104 * Added suppport to allow DB_File to be built with Berkeley DB 3.2 --
105 btree_compare, btree_prefix and hash_cb needed to be changed.
107 * Updated dbinfo to support Berkeley DB 3.2 file format changes.
110 1.74 10th December 2000
112 * A "close" call in DB_File.xs needed parenthesised to stop win32 from
113 thinking it was one of its macros.
115 * Updated dbinfo to support Berkeley DB 3.1 file format changes.
117 * DB_File.pm & the test hasness now use the warnings pragma (when
120 * Included Perl core patch 7703 -- size argument for hash_cb is different
123 * Included Perl core patch 7801 -- Give __getBerkeleyDBInfo the ANSI C
126 * @a = () produced the warning 'Argument "" isn't numeric in entersub'
127 This has been fixed. Thanks to Edward Avis for spotting this bug.
129 * Added note about building under Linux. Included patches.
131 * Included Perl core patch 8068 -- fix for bug 20001013.009
132 When run with warnings enabled "$hash{XX} = undef " produced an
133 "Uninitialized value" warning. This has been fixed.
137 * Added support in version.c for building with threaded Perl.
139 * Berkeley DB 3.1 has reenabled support for null keys. The test
140 harness has been updated to reflect this.
142 1.72 16th January 2000
146 * The module will now use XSLoader when it is available. When it
147 isn't it will use DynaLoader.
149 * The locking section in DB_File.pm has been discredited. Many thanks
150 to David Harris for spotting the underlying problem, contributing
151 the updates to the documentation and writing DB_File::Lock (available
154 1.71 7th September 1999
156 * Fixed a bug that prevented 1.70 from compiling under win32
158 * Updated to support Berkeley DB 3.x
160 * Updated dbinfo for Berkeley DB 3.x file formats.
164 * Initialise $DB_File::db_ver and $DB_File::db_version with
165 GV_ADD|GV_ADDMULT -- bug spotted by Nick Ing-Simmons.
167 * Added a BOOT check to test for equivalent versions of db.h &
172 * fixed a bug in push -- DB_APPEND wasn't working properly.
174 * Fixed the R_SETCURSOR bug introduced in 1.68
176 * Added a new Perl variable $DB_File::db_ver
180 * Merged changes from 5.005_58
182 * Fixed a bug in R_IBEFORE & R_IAFTER procesing in Berkeley DB
185 * Added some of the examples in the POD into the test harness.
189 * Added DBM Filter documentation to DB_File.pm
191 * Fixed DBM Filter code to work with 5.004
193 * A few instances of newSVpvn were used in 1.66. This isn't available in
194 Perl 5.004_04 or earlier. Replaced with newSVpv.
198 * Added DBM Filter code
202 * Fixed a bug in the recno PUSH logic.
203 * The BOOT version check now needs 2.3.4 when using Berkeley DB version 2
205 1.64 21st February 1999
207 * Tidied the 1.x to 2.x flag mapping code.
208 * Added a patch from Mark Kettenis <kettenis@wins.uva.nl> to fix a flag
209 mapping problem with O_RDONLY on the Hurd
210 * Updated the message that db-recno.t prints when tests 51, 53 or 55 fail.
212 1.63 19th December 1998
214 * Fix to allow DB 2.6.x to build with DB_File
215 * Documentation updated to use push,pop etc in the RECNO example &
216 to include the find_dup & del_dup methods.
218 1.62 30th November 1998
220 Added hints/dynixptx.pl.
221 Fixed typemap -- 1.61 used PL_na instead of na
223 1.61 19th November 1998
225 Added a note to README about how to build Berkeley DB 2.x when
227 Minor modifications to get the module to build with DB 2.5.x
228 Fixed a typo in the definition of O_RDONLY, courtesy of Mark Kettenis.
231 Changed the test to check for full tied array support
234 Updated the license section.
236 Berkeley DB 2.4.10 disallows zero length keys. Tests 32 & 42 in
237 db-btree.t and test 27 in db-hash.t failed because of this change.
238 Those tests have been zapped.
240 Added dbinfo to the distribution.
243 Tied Array support was enhanced in Perl 5.004_57. DB_File now
244 supports PUSH,POP,SHIFT,UNSHIFT & STORESIZE.
246 Fixed a problem with the use of sv_setpvn. When the size is
247 specified as 0, it does a strlen on the data. This was ok for DB
248 1.x, but isn't for DB 2.x.
251 If Perl has been compiled with Threads support,the symbol op will be
252 defined. This clashes with a field name in db.h, so it needs to be
253 #undef'ed before db.h is included.
256 Documented the Solaris 2.5 mutex bug
263 Fixed a small bug in the test harness when run under win32
264 The emulation of fd when useing DB 2.x was busted.
268 Added DB_RENUMBER to flags for recno.
272 Patch from Nick Ing-Simmons now allows DB_File to build on NT.
277 Fixed the test harness so that it doesn't expect DB_File to have
278 been installed by the main Perl build.
281 Fixed a bug in mapping 1.x O_RDONLY flag to 2.x DB_RDONLY equivalent
285 DB_File can now build with either DB 1.x or 2.x, but not both at
290 A harmless looking tab was causing Makefile.PL to fail on AIX 3.2.5
292 Small fix for the AIX strict C compiler XLC which doesn't like
293 __attribute__ being defined via proto.h and redefined via db.h. Fix
294 courtesy of Jarkko Hietaniemi.
298 Patch from Gisle Aas <gisle@aas.no> to suppress "use of undefined
299 value" warning with db_get and db_seq.
301 Patch from Gisle Aas <gisle@aas.no> to make DB_File export only the
302 O_* constants from Fcntl.
304 Removed the DESTROY method from the DB_File::HASHINFO module.
306 Previously DB_File hard-wired the class name of any object that it
307 created to "DB_File". This makes sub-classing difficult. Now
308 DB_File creats objects in the namespace of the package it has been
314 Made it illegal to tie an associative array to a RECNO database and
315 an ordinary array to a HASH or BTREE database.
319 Minor changes to DB_FIle.xs and DB_File.pm
323 Documented the incompatibility with version 2 of Berkeley DB.
327 Documented the untie gotcha.
331 Fixed fd method so that it still returns -1 for in-memory files
332 when db 1.86 is used.
336 Minor bug fix in DB_File::HASHINFO, DB_File::RECNOINFO and
339 Changed default mode to 0666.
343 Documented operation of bval.
347 Fixed bug with RECNO, where bval wasn't defaulting to "\n".
351 Minor namespace cleanup: Localized PrintBtree.
355 Made all scripts in the documentation strict and -w clean.
357 Added logic to DB_File.xs to allow the module to be built after
362 Minor documentation changes.
364 Fixed a bug in hash_cb. Patches supplied by Dave Hammen,
365 <hammen@gothamcity.jsc.nasa.govt>.
367 Fixed a bug with the constructors for DB_File::HASHINFO,
368 DB_File::BTREEINFO and DB_File::RECNOINFO. Also tidied up the
369 constructors to make them -w clean.
371 Reworked part of the test harness to be more locale friendly.
375 Documentation update.
377 DB_File now imports the constants (O_RDWR, O_CREAT etc.) from Fcntl
380 The standard hash function exists is now supported.
382 Modified the behavior of get_dup. When it returns an associative
383 array, the value is the count of the number of matching BTREE
388 Merged OS/2 specific code into DB_File.xs
390 Removed some redundant code in DB_File.xs.
392 Documentation update.
394 Allow negative subscripts with RECNO interface.
396 Changed the default flags from O_RDWR to O_CREAT|O_RDWR.
398 The example code which showed how to lock a database needed a call
399 to sync added. Without it the resultant database file was empty.
401 Added get_dup method.
405 Fixed a core dump problem with SunOS.
407 The return value from TIEHASH wasn't set to NULL when dbopen
412 DB_File has been in use for over a year. To reflect that, the
413 version number has been incremented to 1.0.
415 Added complete support for multiple concurrent callbacks.
417 Using the push method on an empty list didn't work properly. This
422 Added prototype support for multiple btree compare callbacks.
426 When DB_File is opening a database file it no longer terminates the
427 process if dbopen returned an error. This allows file protection
428 errors to be caught at run time. Thanks to Judith Grass
429 <grass@cybercash.com> for spotting the bug.