Version 1.21, 2021-09-02
* Crash tolerance
By default it is possible for an abrupt crash (e.g., power failure,
OS kernel panic, or application process crash) to corrupt the gdbm
database file. A new Linux-only mechanism enables applications to
recover the database state corresponding to the most recent
successful gdbm_sync() call before the crash. See the chapter 17
"Crash Tolerance" in the GDBM manual.
* New database file format: numsync
The new "numsync" database format is designed to better support
crash tolerance. To create a database in numsync format, the gdbm_open
(or gdbm_fd_open) function must be given the GDBM_NEWDB|GDBM_NUMSYNC
flags. The GDBM_NUMSYNC flag also takes effect when used together
with GDBM_WRCREAT, provided that the new file is created.
New function gdbm_convert() is provided for converting the databases
from standard GDBM format to numsync and vice versa.
The gdbmtool tool can also be used for converting databases between
these two formats.
* Changes in gdbmtool
** Fix string output in non-ASCII encodings
Printable multi-byte sequences are correctly represented on output.
This also fixes octal representation of unprintable characters.
** The filename variable
This variable supplies the name of database file for use in "open"
command, if the latter is called without arguments. If "open" is
called with the file name argument, the "filename" variable is
initialized to this value.
** The fd variable
If set, its value must be an open file descriptor referring to a
GDBM database file. The "open" command will use gdbm_fd_open
function to use this file. Upon closing the database, this
descriptor will be closed and the variable will be unset.
The file descriptor to use can also be supplied using the
-d (--db-descriptor) command line option.
** The format variable
Defines the format in which new databases will be created. Allowed
values are: "standard" (default) and "numsync".
** New commands: upgrade and downgrade
The "upgrade" command converts current database to the numsync
(extended) format. The "downgrade" command converts current database
to the standard format.
** New command: snapshot
The "snapshot" command is part of the new crash tolerance support.
Given the names of two snapshot files, it analyzes them and selects
the one to be used for database recovery. See the GDBM manual,
section 17.5 "Manual crash recovery" for a detailed discussion of its
use.
Version 1.20, 2021-06-17
* New bucket cache
The bucket cache support has been rewritten from scratch. The new
bucket cache code provides for significant speed up of search
operations.
* Change mmap prereading strategy
Pre-reading of the memory mapper regions, introduced in version 1.19
can be advantageous only when doing intensive look-ups on a read-only
database. It degrades performance otherwise, especially if doing
multiple inserts. Therefore, this version introduces a new flag
to gdbm_open: GDBM_PREREAD. When given, it enables pre-reading of
memory mapped regions.
See https://github.com/Perl/perl5/issues/18884 for details.
Version 1.19 - 2020-12-23
* Pre-read the memory mapped regions on systems that support it.
This speeds up operations on big databases.
* gdbmtool: tagged initialization of structured data
Initializers for structured data can be given in tagged form, e.g.:
store somekey { status=2, id={a,u,x}, name="foo" }
* Bugfixes:
** Preserve locking type during database reorganization
Version 1.18.1 - 2018-10-27
* Fix debian bug 910911
This restores backward compatibility with databases created by version
1.8 (and some later versions, with mmapped I/O disabled).
See https://bugs.debian.org/910911 for a detailed description.
Version 1.18 - 2018-08-21
* Bugfixes:
** Fix directory entry validation
See https://puszcza.gnu.org.ua/bugs/?402
Directory validation function would falsely report corruption after
several directory reallocations.
** Fix improper use of GDBM_DEBUG macro
See https://puszcza.gnu.org.ua/bugs/?401
** Fix spurious error from gdbm_dump and gdbm_export
The functions incorrectly reported as error the GDBM_ITEM_NOT_FOUND
status, which is reported when upon normal termination of iteration
over the database keys.
** Make sure gdbm_sync always returns a meaningful value
See https://puszcza.gnu.org.ua/bugs/?400
Version 1.17 - 2018-07-30
* int gdbm_close and gdbm_sync
Both functions now return 0 on success. On error, they return -1
and set gdbm_errno and errno to the appropriate error codes.
* Fix bug in gdbm_dump function
The function did not clear the GDBM_ITEM_NOT_FOUND error marker, which
is set upon its normal termination.
Version 1.16 - 2018-06-27
* Maintain sorting order of the available block list after coalescing
This is a very long-standing bug, dated back to the time when optional
coalescing of available blocks with the newly released block was
introduced. Merging the released block with an adjacent block of
available space would clobber the sorting order of the available list.
The bug was discovered due to strict database consistency checks,
introduced in version 1.15.
The fix introduced in this version will silently restore the proper
sorting order of available block list before its use.
* Improve block merging algorithm
New implementation of block merging algorithm will correctly handle
both left and right-adjacent blocks.
Version 1.15 - 2018-06-15
* Extensive database consistency checking
GDBM tries to detect inconsistencies in input database files as
early as possible. When an inconcistency is detected, a helpful
diagnostics is returned and the database is marked as needing
recovery. From this moment on, any GDBM function trying to access
the database will immediately return error code (instead of
eventually segfaulting as previous versions did). In order to
reconstruct the database and return it to healthy state, the
gdbm_recover function should be used.
Special thanks to Lionel Debroux and Craig Young for investing
their time and efforts in testing and providing valuable feedback.
* Improved error checking
* Removed gdbm-1.8.3 compatibility layer
* Commands can be given to gdbmtool in the command line
The syntax is:
gdbmtool DBNAME COMMAND [ARGS...]
Multiple commands are separated by semicolon (take care to escape it),
e.g.:
gdbmtool t.db count\; avail
* Fixed data conversion bugs in storing structured keys or content
* New member in the gdbm_recovery structure: duplicate_keys.
Upon return from gdbm_recover, this member holds the number of keys
that has not been recovered, because the same key had already been stored
in the database. The actual number of stored keys is thus
recovered_keys - duplicate_keys.
* New error codes.
GDBM_BAD_BUCKET "Malformed bucket header"
GDBM_BAD_HEADER "Malformed database file header"
GDBM_BAD_AVAIL "Malformed avail_block"
GDBM_BAD_HASH_TABLE "Malformed hash table"
GDBM_BAD_DIR_ENTRY "Invalid directory entry"
pkglint --only "https instead of http" -r -F
With manual adjustments afterwards since pkglint 19.4.4 fixed a few
indentations in unrelated lines.
This mainly affects projects hosted at SourceForce, as well as
freedesktop.org, CTAN and GNU.
Version 1.14.1 - 2018-01-03
* Increment soname current version number.
Version 1.14 - 2018-01-01
* Make sure created databases are byte-for-byte reproducible
This fixes two longstanding bugs: (1) when allocating database file
header blocks, the unused memory is filled with zeroes; (2) when expanding
a mmapped memory area, the added extent is filled with zeroes.
* Fix build with --enable-gdbm-export
* Make gdbm_error global variable thread safe.
* Fix possible segmentation violation in gdbm_setopt
* Fix handling of group headers in --help output.
* gdbm_fetch, gdbm_firstkey, and gdbm_nextkey behavior
If the requested key was not found, these functions return datum with
dptr pointing to NULL and set gdbm_errno to GDBM_ITEM_NOT_FOUND (in
prior releases, gdbm_errno was set to GDBM_NO_ERROR),
If an error occurred, dptr is set to NULL, and gdbm_errno to
the error code.
In any case gdbm_errno is guaranteed to have meaningful value upon
return.
* Error handling
In previous versions of GDBM, fatal errors (such as write error while
storing the key/data pair or bucket) caused immediate termination of
the program via call to exit(3). This is no longer the case.
Starting from this version, if a fatal error occurrs while modifying
the database file, that database is marked as needing recovery and
gdbm_errno is set to GDBM_NEED_RECOVERY. Calls to any GDBM functions,
except gdbm_recover, will then return immediately with the same error
code.
The function gdbm_recover examines the database file and fixes
eventual inconsistencies. Upon successful return it clears the error
state and makes the database operational again.
For backward compatibility, the fatal_func argument to gdbm_open is
retained and its functionality is not changed. If it is not NULL, the
new error handling procedures are disabled, the function it points to
will be called upon fatal errors. When it returns, exit(1) will be
called.
* Per-database error codes
In addition to gdbm_error global variable, the most recent error state
is saved in the GDBM_FILE structure. This facilitates error handling
when operating multiple GDBM databases simultaneously.
* New configuration variable COMPATINCLUDEDIR
When used with --enable-libgdbm-compat, this variable points to the
directory where the headers file dbm.h and ndbm.h will be installed.
Use this variable to avoid conflicts with already installed headers.
E.g.:
./configure --enable-libgdbm-compat COMPATINCLUDEDIR=/usr/include/gdbm
Problems found with existing distfiles:
distfiles/D6.data.ros.gz
distfiles/cstore0.2.tar.gz
distfiles/data4.tar.gz
distfiles/sphinx-2.2.7-release.tar.gz
No changes made to the cstore or mariadb55-client distinfo files.
Otherwise, existing SHA1 digests verified and found to be the same on
the machine holding the existing distfiles (morden). All existing
SHA1 digests retained for now as an audit trail.
problem. Thanks, martin!
Changes:
Version 1.11, 2013-12-25
* Improved dump format.
A new dump format is implemented, which encodes all data in base64 and
stores not only key/data pairs, but also the original database file
metadata, such as file name, mode and ownership. Files in this format
can be sent without additional encapsulation over transmission
channels that normally allow only ASCII data. Dumps in this format
allow for restoring an exact copy of the database, including file
ownership and privileges.
* New function: gdbm_count
int gdbm_count (GDBM_FILE *file, gdbm_count *count);
Counts records in `file' and stores the result in the memory location
pointed to by `count'.
* New utilities: gdbm_dump and gdbm_load.
Gdbm_dump creates a plain-text dump of the GDBM database. This dump
can be used to create an exact copy of the database afterward.
The gdbm_load performs the reverse: given the dump file, it creates a
GDBM database. Apart from native GDBM dump formats, it also understands
the format generated by Berkeley DB db_dump utility. Thus, an easy
way to convert a Berkeley DB database to GDBM is:
db_dump input.db | gdbm_load output.db
* gdbmtool
The gdbmtool utility allows you to examine, modify or create GDBM
databases. It provides an easy-to-use interactive shell and can
be used for scripting. One of the unique features of gdbmtool is that
it allows to define datum structures for key and content parts, similarly
to the C "struct" declarations, and to input and display such
structured data.
* Internationalization
This version of GDBM is fully internationalized. The following
localizations are available: Finnish, German, Japanese, Polish and Ukrainian.
* Support for close-on-exec flag in gdbm_open (see GDBM_CLOEXEC in the docs).
* Improve testgdbm command system
The testgdbm tool now supports multicharacter commands.
* Bugfixes
* Bugfix: Improperly used preprocessor directive caused compilation failure
when using gcc 4.4.4 or newer.
Changes 1.9:
* Use of mmap
* Changes in compatibility mode
* Locking is disabled.
* Do not link pag to dir.
* gdbm_setopt
* The testgdbm program is installed
* A testsuite is provided.
* Improved documentation.
This changes the buildlink3.mk files to use an include guard for the
recursive include. The use of BUILDLINK_DEPTH, BUILDLINK_DEPENDS,
BUILDLINK_PACKAGES and BUILDLINK_ORDER is handled by a single new
variable BUILDLINK_TREE. Each buildlink3.mk file adds a pair of
enter/exit marker, which can be used to reconstruct the tree and
to determine first level includes. Avoiding := for large variables
(BUILDLINK_ORDER) speeds up parse time as += has linear complexity.
The include guard reduces system time by avoiding reading files over and
over again. For complex packages this reduces both %user and %sys time to
half of the former time.
and add a new helper target and script, "show-buildlink3", that outputs
a listing of the buildlink3.mk files included as well as the depth at
which they are included.
For example, "make show-buildlink3" in fonts/Xft2 displays:
zlib
fontconfig
iconv
zlib
freetype2
expat
freetype2
Xrender
renderproto
RECOMMENDED is removed. It becomes ABI_DEPENDS.
BUILDLINK_RECOMMENDED.foo becomes BUILDLINK_ABI_DEPENDS.foo.
BUILDLINK_DEPENDS.foo becomes BUILDLINK_API_DEPENDS.foo.
BUILDLINK_DEPENDS does not change.
IGNORE_RECOMMENDED (which defaulted to "no") becomes USE_ABI_DEPENDS
which defaults to "yes".
Added to obsolete.mk checking for IGNORE_RECOMMENDED.
I did not manually go through and fix any aesthetic tab/spacing issues.
I have tested the above patch on DragonFly building and packaging
subversion and pkglint and their many dependencies.
I have also tested USE_ABI_DEPENDS=no on my NetBSD workstation (where I
have used IGNORE_RECOMMENDED for a long time). I have been an active user
of IGNORE_RECOMMENDED since it was available.
As suggested, I removed the documentation sentences suggesting bumping for
"security" issues.
As discussed on tech-pkg.
I will commit to revbump, pkglint, pkg_install, createbuildlink separately.
Note that if you use wip, it will fail! I will commit to pkgsrc-wip
later (within day).
developer is officially maintaining the package.
The rationale for changing this from "tech-pkg" to "pkgsrc-users" is
that it implies that any user can try to maintain the package (by
submitting patches to the mailing list). Since the folks most likely
to care about the package are the folks that want to use it or are
already using it, this would leverage the energy of users who aren't
developers.
in the process. (More information on tech-pkg.)
Bump PKGREVISION and BUILDLINK_DEPENDS of all packages using libtool and
installing .la files.
Bump PKGREVISION (only) of all packages depending directly on the above
via a buildlink3 include.
All library names listed by *.la files no longer need to be listed
in the PLIST, e.g., instead of:
lib/libfoo.a
lib/libfoo.la
lib/libfoo.so
lib/libfoo.so.0
lib/libfoo.so.0.1
one simply needs:
lib/libfoo.la
and bsd.pkg.mk will automatically ensure that the additional library
names are listed in the installed package +CONTENTS file.
Also make LIBTOOLIZE_PLIST default to "yes".
