SIGHUP,SIGINT,SIGTERM on startup, so make those also conditional to isc_bind9.
The net effect of this was that dhclient blocked the 3 signals for both
itself and its progeny so /etc/rc.d/dhclient restart would not work.
- Finding probable functions
- Replacing instructions
- Emulating instructions
It is tested only on ARMv7 CPUs yet, for example,
-m evbarm (-a earm) kernel=BEAGLEBONE.
* Improvements to autoconf build emulation
* CARRIER/NOCARRIER are now run outside of the IPv4 runs
* validate domains correctly which contain a - on non bash shells
* don't remove IPv6 addresses from internal state when added as tentative
* HUP now rebinds, ALRM now releases - the -x and -k flags work as
they used to
* Add -M, --master option to force dhcpcd into master mode even if one
interface is specified on the command line
* Fix a crash when receiving a reconfigure key
* Dumping a DHCPv4 lease works again
* SEND_DAD code removed
* hoplimit is no longer defined for DHCPv6 messages
* hoplimit of 255 for IPv6 RS/ND message is now defined at the socket
instead of ancillary data with the message
word size definition on sh* to be a compile time constant. We do not
provide a single library supporting both variants, so we pick the
smaller alignment for now.
we still need significant work on these ports:
- mips*
- powerpc
- sh*
- vax
- m68k*
the x86 platforms are probably ready to switch, but i'm not in a good
position to test them. these platforms are probably ready to switch,
as they're still mostly in bringup mode anyway:
- powerpc64
- coldfire
- ia64
- Tidy up header file inclusions
- Replace curcpu with cpu_number()
- Fix argument of dtrace_getreg
- Tweak for gcc 4.8
Note that the file is still not buildable.
- match ARM CPU default of GCC
- better support for SPARC64
- type checkings for abs/fabs
- null pointer checks against struct member
- various other improvements
and netbsd-elf.h on powerpc64-netbsd.
port netbsd64.h forward from GCC changes. (it's a pity that
a vast portion of linux64.h isn't in some common header that
netbsd64.h and freebsd64.h could also used. there's 100s of
lines of copypasta here, oh well.)
Sigh; one more attempt. This time I'm sure I've verified that the
.pc files contain the right number and that atf-version also outputs
the right stuff... Both with a clean and non-clean obj directory.
Should fix part of the problems reported in PR bin/48624.
- set GNUHOSTDIST
- move gthr-default.h to BUILDINCS, and add glue to add it to COPYHEADERS
- adjust COPYHEADERS to cope with files outside of dist/libstdc++-v3/.
# XXX these rules don't always work if the ${.TARGET} ends up being the
# copy in ../../arch/$arch/ and that version is older. (but will
# only break read-only source builds.)
highlights from: http://gcc.gnu.org/gcc-4.6/changes.html
GCC now has stricter checks for invalid command-line options
New -Wunused-but-set-variable and -Wunused-but-set-parameter
warnings
Many platforms have been obsoleted
Link-time optimization improvements
A new switch -fstack-usage has been added
A new function attribute leaf was introduced
A new warning, enabled by -Wdouble-promotion
Support for selectively enabling and disabling warnings via
#pragma GCC diagnostic has been added
There is now experimental support for some features from the
upcoming C1X revision of the ISO C standard
Improved experimental support for the upcoming C++0x ISO C++
standard
G++ now issues clearer diagnostics in several cases
Updates for ARM, x86, MIPS, PPC/PPC64, SPARC
Darwin, FreeBSD, Solaris 2, MinGW and Cygwin now all support
__float128 on 32-bit and 64-bit x86 targets. [*1]
highlights from: http://gcc.gnu.org/gcc-4.7/changes.html
The -fconserve-space flag has been deprecated
Support for a new parameter --param case-values-threshold=n
was added
Interprocedural and Link-time optimization improvements
A new built-in, __builtin_assume_aligned, has been added
A new warning option -Wunused-local-typedefs was added
A new experimental command-line option -ftrack-macro-expansion
was added
Support for atomic operations specifying the C++11/C11 memory
model has been added
There is support for some more features from the C11 revision
of the ISO C standard
Improved experimental support for the new ISO C++ standard,
C++11
Updates for ARM, x86, MIPS, PPC/PPC64, SH, SPARC, TILE*
A new option (-grecord-gcc-switches) was added
highlights from: http://gcc.gnu.org/gcc-4.8/changes.html
GCC now uses C++ as its implementation language. This means
that to build GCC from sources, you will need a C++
compiler that understands C++ 2003
DWARF4 is now the default when generating DWARF debug
information
A new general optimization level, -Og, has been introduced
A new option -ftree-partial-pre was added
The option -fconserve-space has been removed
The command-line options -fipa-struct-reorg and
-fipa-matrix-reorg have been removed
Interprocedural and Link-time optimization improvements
AddressSanitizer, a fast memory error detector, has been
added [*2]
A new -Wsizeof-pointer-memaccess warning has been added
G++ now supports a -std=c++1y option for experimentation
with features proposed for the next revision of the
standard, expected around 2014
Improved experimental support for the new ISO C++ standard,
C++11
A new port has been added to support AArch64
Updates for ARM, x86, MIPS, PPC/PPC64, SH, SPARC, TILE*
[*1] we should support this too!
[*2] we should look into this.
https://code.google.com/p/address-sanitizer/
BIND 9.10.0b1 is the first beta development release of BIND 9.10,
a new branch of BIND 9.
This document summarizes features added or significantly changed
since the previous major release, BIND 9.9. Items that were not
in the previous development release, BIND 9.10.0a2, are marked
with asterisks (**). Bug fixes since the previous development
release are also summarized.
Please see the CHANGES file in the source code release for a
complete list of all changes.
Download
The latest versions of BIND 9 software can always be found on
our web site at http://www.isc.org/downloads/. There you will
find additional information about each release, source code, and
pre-compiled versions for Microsoft Windows operating systems.
Support
Professional support is provided by Internet Systems Consortium,
Inc., doing business as DNSco. Information about paid support
options is available at http://www.dns-co.com/solutions/. Free
support is provided by our user community via a mailing list.
Information on all public email lists is available at
https://www.isc.org/community/mailing-list/.
New Features
DNS Response-rate limiting (DNS RRL), which blunts the impact
of reflection and amplification attacks, is always compiled in
and no longer requires a compile-time option to enable it.
An experimental "Source Identity Token" (SIT) EDNS option is now
available. Similar to DNS Cookies (as invented by Donald Eastlake
III and described in draft-eastlake-dnsext-cookies-04), these
are designed to enable clients to detect off-path spoofed
responses, and to enable servers to detect spoofed-source queries.
Servers can be configured to send smaller responses to clients
that have not identified themselves using a SIT option, reducing
the effectiveness of amplification attacks. RRL processing has
also been updated: clients proven to be legitimate via SIT are
not subject to rate limiting. Use "configure --enable-sit" to
enable this feature in BIND 9. [RT #35389] **
A new zone file format, "map", stores zone data in a format that
can be mapped directly into memory, allowing significantly faster
zone loading. [RT #25419]
"delve" (domain entity lookup and validation engine) is a new
tool with dig-like semantics for looking up DNS data and performing
internal DNSSEC validation. This allows easy validation in
environments where the resolver may not be trustworthy, and
assists with troubleshooting of DNSSEC problems. (Note: not yet
available on Windows.) [RT #32406] **
The new "prefetch" option can improve recursive resolver
performance: when it is in use, cache records that are still
being requested by clients will automatically be refreshed from
the authoritative server before they expire, reducing or eliminating
the time window in which no answer is available in the cache.
[RT #35041]
Improved EDNS processing allows better resolver performance and
reliability over slow or lossy connections. [RT #30655]
Substantial improvements have been made in response-policy zone
(RPZ) performance. Up to 32 response-policy zones can now be
configured. Performance loss due to adding additional RPZs is
minimal.
RPZ now allows response policies to be configured based on the
IP address of the client.
ACLs can now be specified based on geographic location using the
MaxMind GeoIP databases. Use "configure --with-geoip" to enable
this feature in BIND 9. Thanks to Ken Brownfield for the
contribution. [RT #30681]
The version 3 XML schema for the statistics channel, including
new statistics and a flattened XML tree for faster parsing, is
no longer optional. The version 2 XML schema is now deprecated.
[RT #30023]
Improvements have been made to the XSL stylesheet used for XML
statistics: The stylesheet can now be cached by the browser;
section headers are omitted when the sections have no data to
display; counter readability has been improved. Also, broken-out
subgroups of XML statistics (server, zones, net, tasks, mem, and
status) can now be requested. Thanks to Timothe Litt for the
assistance. [RT #35115] [RT #35117]
The statistics channel can now provide data in JSON format as
well as XML.
Per-zone stats counters have been added to track TCP and UDP
queries. [RT #35375] **
Server-wide stats counters have been added to track EDNS options
received. [RT #35447] **
The new "in-view" zone option allows zone data to be shared
between views, so that multiple views can serve the same zones
authoritatively without storing multiple copies in memory. [RT #32968]
A new compile-time option, "configure --enable-native-pkcs11",
allows the BIND 9 cryptography functions to use the PKCS#11 API
natively, so that BIND can drive a cryptographic hardware service
module (HSM) directly instead of using a modified OpenSSL as an
intermediary. This has been tested with the Thales nShield HSM
and with SoftHSMv2 from the OpenDNSSEC project. [RT #29031]
When re-signing a zone, the new "dnssec-signzone -Q" option drops
signatures from keys that are still published but are no longer
active. Thanks to Pierre Beyssac for the contribution. [RT #34990]
New options have been added to "dnssec-coverage": -z and -k
indicate whether to limit coverage checks to ZSK's or KSK's, and
-l limits coverage checking to a specified duration. Thanks to
Peter Palfrader for the contribution. [RT #35168]
"named-checkconf -px" will print the contents of configuration
files with the shared secrets obscured, making it easier to share
configuration (e.g. when submitting a bug report) without revealing
private information. [RT #34465]
Added a "no-case-compress" ACL, which causes "named" to use
case-insensitive compression for specified clients. This is
useful when dealing with broken client implementations that use
case-sensitive name comparisons, rejecting responses that fail
to match the capitalization of the query that was sent.
"named" now preserves the capitalization of names when responding
to queries: for instance, a query for "example.com" may be
answered with "example.COM" if the name was configured that way
in the zone file. Some clients have a bug causing them to depend
on the older behavior, in which the case of the answer always
matched the case of the query, rather than the case of the name
configured in the DNS. Such clients can now be specified in the
new "no-case-compress" ACL; this will restore the older behavior
of "named" for those clients only. [RT #35300] **
On operating systems that support routing sockets, including Mac
OSX, *BSD and Linux, network interfaces are re-scanned automatically
whenever they change. Use "automatic-interface-scan no;" to
disable this feature. [RT #23027] **
Added "rndc scan" to trigger an interface scan manually. [RT #23027] **
A new compile-time option, "configure --with-tuning=3Dlarge", tunes
various compiled-in constants and default settings to values
suited to large servers with abundant memory. This can improve
performance on such servers, but will consume more memory and
may degrade performance on smaller systems. [RT #29538] **
The new "max-zone-ttl" option enforces maximum TTLs for zones.
If loading a zone containing a higher TTL, the load fails. DDNS
updates with higher TTLs are accepted but the TTL is truncated.
(Note: Currently supported for master zones only; inline-signing
slaves will be added.) [RT #38405] **
Added a new "dig +subnet" option to send an EDNS CLIENT-SUBNET
option (as described in draft-vandergaast-edns-client-subnet-02)
containing the specified address/prefix when querying. Thanks
to Wilmer van der Gaast for the contribution. [RT #35415] **
Partially implemented the EDNS EXPIRE option (as described in
draft-andrews-dnsext-expire-00). "dig +expire" sends an EXPIRE
option when querying. When this option is sent with an SOA query
to a slave zone running on a server that supports the option,
the response will report the time until the slave zone expires.
EXPIRE uses an experimental option code (65002), which is subject
to change when a permanent code is assigned by IANA. [RT #35416] **
Multiple DLZ databases can now be configured, and are searched
in order to find one that can answer an incoming query. Individual
zones can now be configured to be served from a specific DLZ
database. DLZ databases can serve zones of type "master" and
"redirect".
"named-checkzone" and "named-compilezone" can now read journal
files, allowing them to process dynamic zones without the zones
needing to be frozen first.
The "rndc" command now supports new key algorithms in addition
to HMAC-MD5, including HMAC-SHA1, -SHA224, -SHA256, -SHA384, and
-SHA512. The -A option to rndc-confgen can be used to select
the algorithm for the generated key. (The default is still
HMAC-MD5; this may change in a future release.) [RT #20363]
The internal and export versions of the BIND libraries (libisc,
libdns, etc) have been unified so that external library clients
can use the same libraries as BIND itself. [RT #33131]
Added a "Configure" script for Windows to simplify enabling or
disabling optional features. All versions of Visual Studio up
to 2013 are now supported, and support has been added for 64-bit
builds. Zip files containing pre-compiled 64-bit versions of
BIND 9 are now included with releases. [RT #34160] **
"rndc zonestatus" reports information about a specified zone.
"named" now listens on IPv6 as well as IPv4 interfaces by default.
Feature Changes
The default setting for the -U option (setting the number of UDP
listeners per interface) has been adjusted to improve performance.
[RT #35417] **
Updated zkt and nslint in the contrib directory to the newest
versions: zkt 1.1.2 and nslint-3.0a2. **
The isc_bitstring API is no longer used and has been removed
from the libisc library. [RT #35284] **
The word "never" can now be used as a synonym for "none" when
configuring key event dates in the dnssec tools. [RT #35277]
**
The new libiscpk11 library, introduced in the previous development
release to support native PKCS#11, has been merged into libisc
to simplify dependencies. [RT #35205] **
Documentation of native PKCS#11 has been expanded, specifically
to describe the new pkcs11: URI format used in key labels. [RT #35287] *=
*
The Windows installer now places files in the Program Files area
rather than system services. [RT #35361] **
The timestamps included in RRSIG records can now be read as
integers indicating the number of seconds since the UNIX epoch,
in addition to being read as formatted dates in YYYYMMDDHHMMSS
format. [RT #35185]
The irs_resconf_load() function, used for reading /etc/resolv.conf,
now returns ISC_R_FILENOTFOUND when the file is missing or
unreadable. However, it will still initialize an irs_resconf
structure as if the file had been configured with nameservers
at the IPv4 and IPv6 localhost addresses. Existing code that
uses irs_resconf_load() will need to be updated to treat
ISC_R_FILENOTFOUND as a qualified success, or it may leak memory
due to treating the result as a failure even though an irs_resconf
structure was allocated; see CHANGES for sample C code that
implements the correct behavior [RT #35194]
Bug Fixes
"dnssec-keygen" could set the publication date incorrectly when
only the activation date was specified on the command line. [RT #35278]
Fixed a type mismatch causing the ODBC DLZ driver to dump core
on 64-bit systems. [RT #35324]
Improved building with libtool. [RT #35314]
When a server is specified by name in "nsupdate", all addresses
for that name will be tried before giving up. Previously, if
the first address for the server name was not reachable the
update would fail. [RT #25784]
Fixed an assertion failure caused by using "rndc retransfer"
with inline-signing zones. [RT #35353]
Fixed a build failure from using "./configure --enable-openssl-hash".
[RT #35343]
The "delegation-only" flag now works in zones of type "forward".
(This had previously been documented to work, but this was
actually rejected by the configuration parser.) [RT #35392]
Fixed a race condition which could lead to a core dump when
destroying a resolver fetch object. [RT #35385]
Addressed a potential REQUIRE failure that could occur when
printing out an rdataset using a format that includes comment
data.
The "allow-notify" ACL formerly ignored TSIG keys; this has been
corrected. [RT #35425]
Fixed an uninitialized pointer in log.c that could potentially
have caused a core dump on some platforms. [RT #35260]
Thank You
Thank you to everyone who assisted us in making this release
possible. If you would like to contribute to ISC to assist us
in continuing to make quality open source software, please visit
our donations page at http://www.isc.org/donate/.
(c) 2001-2014 Internet Systems Consortium
- convert to using ${EXTERNAL_GCC_SUBDIR}
- define base-external-gpl3-gcc* subdir as GCC_SUBDIR
- use <bsd.init.mk> over <bsd.own.mk> for a bunch of places; mostly
because it arranges for ../Makefile.inc to be included earlier, and
don't bother including the latter if the former is already included.
- move all .PATH: settings after <bsd.{own,lib}.mk> so that all
valid variables are set before it is evaluated
- rename mknative-gcc* to match their subdir name.
XXX the relationship between the Makefile.inc/Makefile.gcc_path files
is kind of sketchy, it would be great if this was fixed.
upgrading the normal GCC to 4.8.
this tree has had ChangeLog entries removed, as well as all the
other components we delete, and "only" weighs in around 140MB now.
* IAID must be inside an interface block.
* Detect IPv6 address flags on Linux.
* Check that we have ctrl_interface defined in wpa_supplicant.conf and pass this parameter to wpa_cli(8). If not set, warn about not interacting with wpa_supplicant(8).
* Skip Virtual Interface Masters. Currently this only applies to FreeBSD VAP masters as dhcpcd should only work on wlandev created clones.
* Fix handling of ARP failures, thanks to Sebastian Huber.
* Implement RFC 1321 MD5 Message-Digest if not provided in libc.
* Implement RFC 2104 HMAC Keyed Hashing.
* Implement RFC 3118 Authentication for DHCP Messages and RFC 3315 Authentication options.
* Add nodhcp and nodhcp6 directives. Thanks to Sebastian Huber for the initial patch and testing.
* Implement support for RFC 3203, FORCERENEW message.
* Implement Force Renew Nonce Capability option, RFC6704
* Support RECONFIGURE DHCPv6 messages.
* Add support for DHCP auto configuration, RFC 2563.
* Add support for the following RFCs
+ DHCP SLP Directory Agent, RFC2610
+ DHCP Name Service Search, RFC2937
+ DHCP PANA Authentication Agent, RFC5192
+ DHCP Lost Server, RFC5223
+ DHCP CAPWAP, RFC5417
+ DHCP Mobility Services, RFC5678
+ DHCP SIP UA, RFC6011
+ DHCP ANDSF, RFC6153
+ DHCP RDNSS Selection for MIF Nodes, RFC6731
+ DHCP TFTP Server Address, RFC5859
+ DHCP PXELINUX, RFC5071
+ DHCP Access Network Domain Name, RFC5986
+ DHCP Virtual Subnet Selection, RFC6607
+ DHCP Relay Agent Remote-ID, RFC4649
+ DHCP Relay Agent Subscriber-ID, RFC4580
+ DHCPv6 Relay-ID, RFC5460
+ DHCPv6 LIS Discovery, RFC5986
+ DHCPv6 SIP UA, RFC6011
+ DHCPv6 Network Boot, RFC5970
+ DHCPv6 Home Info Discovery in MIPv6, RFC6610
+ DHCPv6 RDNSS Selection for MIF Nodes, RFC6731
+ DHCPv6 Kerberos, RFC6784
+ DHCPv6 Relay-Triggered Reconfiguration, RFC6977
+ DHCPv6 SOL_MAX_RT, RFC7083
* Open UDP sockets for *.*.*.*:bootpc and $ip_address:bootpc to avoid kernel ICMP unreachable messages
* Moved global variables into context variables so dhcpcd is entirely thread safe.
dhcpcd doens't use threads, but it can now be used in a pure threads (ie no process) environment.
* Remove DEBUG_MEMORY guard and always free memory and resources. Remove all atexit(3) and exit(3) calls, instead exiting via the eloop.
* Replace get_line with sscanf where applicable.
* Remove custom set_cloexec and set_nonblock functions. Instead pass O_CLOEXEC or SOCK_CLOEXEC to open, socket, etc.
* Don't use PATH_MAX - we already know the maximum length from the initial directory and IF_NAMESIZE.
* Add a USE_SIGNALS define. If we're not using signals then we also create a control socket in the non MASTER case. We then use this to communicate instead of signals.
* Fix arping moving to the next test if we find the IP address but do not have a profile for it. Thanks to David McGurty for the analysis.
* Respect initial commandline arguments when rebooting via a signal.
2014-02-11 - Release 3.8.3.1
SQLite version 3.8.3.1 fixes a bug present in versions 3.8.1, 3.8.2 and 3.8.3 that can cause queries to omit valid out rows. Upgrading from those versions is recommended.
The problem only comes up if SQLite is compiled with either the SQLITE_ENABLE_STAT3 or SQLITE_ENABLE_STAT4 compile-time options. In that case, if a query has a WHERE clause that contains expressions like this:
WHERE (expr1 OR expr2 OR ... OR exprN) AND column IS NOT NULL
Where all of expr1 through exprN are suitable for use by indexes, then during query planning SQLite might mistakenly converted the "column IS NOT NULL" term into "column>NULL". But the latter term is never true, and so the query would return no rows.
The trouble ticket for this bug is [4c86b126f2]. It is recommended that all users upgrade to avoid this problem.
2014-02-03 - Release 3.8.3
SQLite version 3.8.3 is a regularly scheduled maintenance release. Upgrading from the previous release is optional.
The most visible change in version 3.8.3 is the addition of support for common table expressions. It is now possible to write a single SELECT statement that will query a tree or graph, using either a depth-first or a breadth-first search. A single SQLite query will even solve Sudoku puzzles or compute the Mandelbrot set. As part of this change, SQLite now accepts a VALUES clause anyplace that a SELECT statement is valid.
This release also includes many small performance enhancements which should give a small speed boost to legacy applications. And there are other minor enhancements such as the addition of the printf() SQL function. See the change log for details.
2013-12-06 - Release 3.8.2
SQLite version 3.8.2 is a regularly scheduled maintenance release. Upgrading from the previous release is optional.
Version 3.8.2 adds support for WITHOUT ROWID tables. This is a significant extension to SQLite. Database files that contain WITHOUT ROWID tables are not readable or writable by prior versions of SQLite, however databases that do not use WITHOUT ROWID tables are fully backwards and forwards compatible.
The 3.8.2 release contains a potentially incompatible change. In all prior versions of SQLite, a cast from a very large positive floating point number into an integer resulted in the most negative integer. In other words, CAST(+99.9e99 to INT) would yield -9223372036854775808. This behavior came about because it is what x86/x64 hardware does for the equivalent cast in the C language. But the behavior is bizarre. And so it has been changed effective with this release so that a cast from a floating point number into an integer returns the integer between the floating point value and zero that is closest to the floating point value. Hence, CAST(+99.9e99 to INT) now returns +9223372036854775807. Since routines like sqlite3_column_int64() do an implicit cast if the value being accessed is really a floating point number, they are also affected by this change.
Besides the two changes mentioned above, the 3.8.2 release also includes a number of performance enhancements. The skip-scan optimization is now available for databases that have been processed by ANALYZE. Constant SQL functions are now factored out of inner loops, which can result in a significant speedup for queries that contain WHERE clause terms like "date>datetime('now','-2 days')". And various high-runner internal routines have been refactored for reduced CPU load.
2013-10-17 - Release 3.8.1
SQLite version 3.8.1 is a regularly scheduled maintenance release. Upgrading from the previous release is optional, though you should upgrade if you are using partial indices as there was a bug related to partial indices in the previous release that could result in an incorrect answer for count(*) queries.
The next generation query planner that was premiered in the previous release continues to work well. The new query planner has been tweaked slightly in the current release to help it make better decisions in some cases, but is largely unchanged. Two new SQL functions, likelihood() and unlikely(), have been added to allow developers to give hints to the query planner without forcing the query planner into a particular decision.
Version 3.8.1 is the first SQLite release to take into account the estimated size of table and index rows when choosing a query plan. Row size estimates are based on the declared datatypes of columns. For example, a column of type VARCHAR(1000) is assumed to use much more space than a column of type INT. The datatype-based row size estimate can be overridden by appending a term of the form "sz=NNN" (where NNN is the average row size in bytes) to the end of the sqlite_stat1.stat record for a table or index. Currently, row sizes are only used to help the query planner choose between a table or one of its indices when doing a table scan or a count(*) operation, though future releases are likely to use the estimated row size in other contexts as well. The new PRAGMA stats statement can be used to view row size estimates.
Version 3.8.1 adds the SQLITE_ENABLE_STAT4 compile-time option. STAT4 is very similar to STAT3 in that it uses samples from indices to try to guess how many rows of the index will be satisfy by WHERE clause constraints. The difference is that STAT4 samples all columns of the index whereas the older STAT3 only sampled the left-most column. Users of STAT3 are encouraged to upgrade to STAT4. Application developers should use STAT3 and STAT4 with caution since both options, by design, violate the query planner stability guarantee, making it more difficult to ensure uniform performance is widely-deployed and mass-produced embedded applications.
2013-09-03 - Release 3.8.0.2
SQLite version 3.8.0.2 contains a one-line fix to a bug in the new optimization that tries to omit unused LEFT JOINs from a query.
2013-08-29 - Release 3.8.0.1
SQLite version 3.8.0.1 fixes some obscure bugs that were uncovered by users in the 3.8.0 release. Changes from 3.8.0 are minimal.
2013-08-26 - Release 3.8.0
Do not fear the zero!
SQLite version 3.8.0 might easily have been called "3.7.18" instead. However, this release features the cutover of the next generation query planner or NGQP, and there is a small chance of breaking legacy programs that rely on undefined behavior in previous SQLite releases, and so the minor version number was incremented for that reason. But the risks are low and there is a query planner checklist is available to application developers to aid in avoiding problems.
SQLite version 3.8.0 is actually one of the most heavily tested SQLite releases ever. Thousands and thousands of beta copies have be downloaded, and presumably tested, and there have been no problem reports.
In addition to the next generation query planner, the 3.8.0 release adds support for partial indices, as well as several other new features. See the change log for further detail.
2013-05-20 - Release 3.7.17
SQLite version 3.7.17 is a regularly schedule maintenance release. Visit the change log for a full explanation of the changes in this release.
There are many bug fixes in version 3.7.17. But this does not indicate that 3.7.16 was a problematic release. All of the bugs in 3.7.17 are obscure and are unlikely to impact any particular application. And most of the bugs that are fixed in 3.7.17 predate 3.7.16 and have been in the code for years without ever before being noticed. Nevertheless, due to the large number of fixes, all users are encouraged to upgrade when possible.
2013-04-12 - Release 3.7.16.2
SQLite version 3.7.16.2 fixes a long-standing flaw in the Windows OS interface that can result in database corruption under a rare race condition. See http://www.sqlite.org/src/info/7ff3120e4f for a full description of the problem.
As far as we know, this bug has never been seen in the wild. The problem was discovered by the SQLite developers while writing stress tests for a separate component of SQLite. Those stress tests have not yet found any problems with the component they were intended to verify, but they did find the bug which is the subject of this patch release.
Other than updates to version numbers, the only difference between this release and 3.7.16.1 is a two-character change in a single identifier, which is contained in the windows-specific OS interface logic. There are no changes in this release (other than version numbers) for platforms other than Windows.
2013-03-29 - Release 3.7.16.1
SQLite version 3.7.16.1 is a bug fix release that fixes a few problems that were present in the previous releases.
The primary motivation for version 3.7.16.1 is to fix a bug in the query optimizer that was introduced as part of version 3.7.15. The query optimizer was being a little overzealous in optimizing out some ORDER BY clauses, which resulted in sorting being omitted on occasions where sorting is required to get the correct answer. See ticket a179fe7465 for details.
In addition to the ORDER BY fix, several other patches to fix obscure (and mostly harmless) bugs and to fix spelling errors in source code comments are also included in this release.
2013-03-18 - Release 3.7.16
SQLite version 3.7.16 is a regularly scheduled release of SQLite. This release contains several language enhancements and improvements to the query optimizer. A list of the major enhancements and optimizations can be see on the change log.
There was one important bug fix (see Ticket fc7bd6358f) that addresses an incorrect query result that could have occurred in a three-way join where the join constraints compared INTEGER columns to TEXT columns. This issue had been in the code for time out of mind and had never before been reported, so we surmise that it is very obscure. Nevertheless, all users are advised to upgrade to avoid any future problems associated with this issue.
2013-01-09 - Release 3.7.15.2
SQLite version 3.7.15.2 is a patch release that fixes a single bug that was introduced in version version 3.7.15. The fix is a 4-character edit to a single line of code. Other than this 4-character change and the update of the version number, nothing has changed from version 3.7.15.1.
2012-12-19 - Release 3.7.15.1
SQLite version 3.7.15.1 is a patch release that fixes a single bug that was introduced in version version 3.7.15. The fix involved changing two lines of code and adding a single assert(). This release also includes some new test cases to prevent a regression of the bug, and the version number is increased, of course. But otherwise, nothing has changed from version 3.7.15.
2012-12-12 - Release 3.7.15
SQLite version 3.7.15 is a regularly schedule release of SQLite. This release contains several improvements to the query planner and optimizer and one important bug fix. This is the first release to officially support Windows 8 Phone.
The important bug fix is a problem that can lead to segfaults when using shared cache mode on a schema that contains a COLLATE operator within a CHECK constraint or within a view. Collating functions are associated with individual database connections. But a pointer to the collating function was also being cached within expressions. If an expression was part of the schema and contained a cached collating function, it would point to the collating function in the database connection that originally parsed the schema. If that database connection closed while other database connections using the same shared cache continued to operate, they other database connections would try to use the deallocated collating function in the database connection that closed. The fix in version 3.7.15 was to not cache collating function pointers in the expression structure but instead look them up each time a new statement is prepared.
This release also contains some important enhancements to the query planner which should (we hope) make some queries run faster. The enhancements include:
When doing a full-table scan, try to use an index instead of the original table, under the theory that indices contain less information and are thus smaller and hence require less disk I/O to scan.
Enhance the IN operator to allow it to make use of indices that have numeric affinity.
Do a better job of recognizing when an ORDER BY clause can be implemented using indices - especially in cases where the ORDER BY clause contains terms from two or more tables in a join.
2012-10-04 - Release 3.7.14.1
SQLite version 3.7.14.1 is a patch release. Changes from the baseline version 3.7.14 are minimal and are restricted to fixing three bugs.
One of the fixed bugs is a long-standing issue with the TCL interface. Another is an external compiler bug that SQLite merely works around and that only comes up if you are using the VisualStudio-2012 compiler to generate WinRT applications on ARM with optimizations enabled. The third problem is an SQLite core bug, introduced in version 3.7.14, that can cause a segfault if a query contains a LEFT JOIN that contains an OR in the ON clause.
2012-09-03 - Release 3.7.14
SQLite version 3.7.14 is a regularly scheduled maintenance release of SQLite. The previous release continues to work well. Upgrading is optional.
Version 3.7.14 drops native support for OS/2. We are not aware of any active projects that were using SQLite on OS/2 and since the SQLite developers had no way of testing on OS/2 it seemed like it was time to simply remove the OS/2 code from the SQLite tree. If there are OS/2 projects out there that still need SQLite support, they can continue to maintain their own private VFS which can be linked to SQLite at start-time using the sqlite3_vfs_register() interface.
The sqlite3_close_v2() interface has been added. The sqlite3_close_v2() interface differs from sqlite3_close() in that it is designed to work better for host language that use a garbage collector. With the older sqlite3_close() interface, the associated prepared statements and sqlite3_backup objects must be destroyed before the database connection. With the newer sqlite3_close_v2() interface, the objects can be destroyed in any order.
This release also includes performance improvements to the sort algorithm that is used to implement ORDER BY and CREATE INDEX. And the query planner has been enhanced to better use covering indices on queries that use OR terms in the WHERE clause.
2012-06-11 - Release 3.7.13
SQLite version 3.7.13 adds support for WinRT and metro style applications for Microsoft Windows 8. The 3.7.13 release is coming sooner than is usual after the previous release in order to get this new capability into the hands of developers. To use SQLite in a metro style application, compile with the -DSQLITE_OS_WINRT flag. Because of the increased application security and safety requirements of WinRT, all database filenames should be full pathnames. Note that SQLite is not capable of accessing databases outside the installation directory and application data directory. This restriction is another security and safety feature of WinRT. Apart from these restrictions, SQLite should work exactly the same on WinRT as it does on every other system.
Also in this release: when a database is opened using URI filenames and the mode=memory query parameter then the database is an in-memory database, just as if it had been named ":memory:". But, if shared cache mode is enabled, then all other database connections that specify the same URI filename will connect to the same in-memory database. This allows two or more database connections (in the same process) to share the same in-memory database.
This release also includes some corner-case performance optimizations that are obscure yet significant to an important subset of SQLite users. Getting these performance optimizations into circulation quickly is yet another reason for making this release so soon following the previous.
The next release of SQLite is scheduled to occur after the usual 2 or 3 month interval.
2012-05-22 - Patch Release 3.7.12.1
SQLite version 3.7.12.1 is a patch release for version 3.7.12 that fixes a bug that was introduced in version 3.7.12 and that can cause a segfault for certain obscure nested aggregate queries. There are very few changes in 3.7.12.1, and upgrading is only needed for applications that do nested aggregate queries.
2012-05-14 - Version 3.7.12
SQLite version 3.7.12 is a regularly scheduled maintenance release. This release contains several new optimizations and bug fixes and upgrading is recommended. See the change summary for details.
2012-03-20 - Version 3.7.11
SQLite version 3.7.11 is a regularly scheduled maintenance release which was rushed out early due to a bug in the query optimizer introduced in the previous release. The bug is obscure - it changes a LEFT JOIN into an INNER JOIN in some cases when there is a 3-way join and OR terms in the WHERE clause. But it was considered serious enough to rush out a fix. Apart from this one problem, SQLite version 3.7.10 has not given any trouble. Upgrading to version 3.7.11 from versions 3.7.6.3, 3.7.7, 3.7.7.1, 3.7.8, or 3.7.9 is optional. Upgrading from other releases, including the previous release 3.7.10, is recommended.
Other enhancements found in this release are enumerated in the change log.
2012-01-16 - Version 3.7.10
SQLite version 3.7.10 is a regularly scheduled maintenance release. Upgrading from version 3.7.6.3, 3.7.7, 3.7.7.1, 3.7.8, or 3.7.9 is optional. Upgrading from other releases is recommended.
The SQLITE_CONFIG_PCACHE mechanism has been replaced with SQLITE_CONFIG_PCACHE2. If you do not know what this mechanism is (it is an extreme corner-case and is seldom used) then this change will not effect you in the least.
The default schema format number for new database files has changed from 1 to 4. SQLite has been able to generate and read database files using schema format 4 for six years. But up unto now, the default schema format has been 1 so that older versions of SQLite could read and write databases generated by newer versions of SQLite. But those older versions of SQLite have become so scarce now that it seems reasonable to make the new format the default.
SQLite is changing some of the assumptions it makes above the behavior of disk drives and flash memory devices during a sudden power loss. This change is completely transparent to applications. Read about the powersafe overwrite property for additional information.
Lots of new interfaces have been added in this release:
sqlite3_db_release_memory()
PRAGMA shrink_memory
sqlite3_db_filename()
sqlite3_stmt_busy()
sqlite3_uri_boolean()
sqlite3_uri_int64()
The PRAGMA cache_size statement has been enhanced. Formerly, you would use this statement to tell SQLite how many pages of the database files it should hold in its cache at once. The total memory requirement would depend on the database page size. Now, if you give PRAGMA cache_size a negative value -N, it will allocate roughly N kibibytes of memory to cache, divided up according to page size. This enhancement allows programs to more easily control their memory usage.
There have been several obscure bug fixes. One noteworthy bug, ticket ff5be73dee, could in theory result in a corrupt database file if a power loss occurred at just the wrong moment on an unusually cantankerous disk drive. But that is mostly a theoretical concern and is very unlikely to happen in practice. The bug was found during laboratory testing and has never been observed to occur in the wild.
2011-11-01 - Version 3.7.9
SQLite version 3.7.9 is a regularly scheduled maintenance release. Upgrading from version 3.7.6.3, 3.7.7, 3.7.7.1, and 3.7.8 is optional. Upgrading from other versions is recommended.
The SQLITE_ENABLE_STAT2 compile-time option is now a no-op. The enhanced query-planner functionality formerly available using SQLITE_ENABLE_STAT2 is now available through SQLITE_ENABLE_STAT3. The enhanced query planning is still disabled by default. However, future releases of SQLite might convert STAT3 from an enable-option to a disable-option so that it is available by default and is only omitted upon request.
The FTS4 full-text search engine has been enhanced such that tokens in the search string that begin with "^" must be the first token in their respective columns in order to match. Formerly, "^" characters in the search string were simply ignored. Hence, if a legacy application was including "^" characters in FTS4 search strings, thinking that they would always be ignored, then those legacy applications might break with this update. The fix is simply remove the "^" characters from the search string.
See the change summary for additional changes associated with this release.
2011-September-19 - Version 3.7.8
SQLite version 3.7.8 is a quarterly maintenance release. Upgrading from versions 3.7.6.3, 3.7.7, or 3.7.7.1 is optional. Upgrading from other versions is recommended.
This release features a new "external merge sort" algorithm used to implement ORDER BY and GROUP BY and also to presort the content of an index for CREATE INDEX. The new algorithm does approximately the same number of comparisons and I/Os as before, but the I/Os are much more sequential and so runtimes are greatly reduced when the size of the set being sorted is larger than the filesystem cache. The performance improvement can be dramatic - orders of magnitude faster for large CREATE INDEX commands. On the other hand, the code is slightly slower (1% or 2%) for a small CREATE INDEX. Since CREATE INDEX is not an operation that commonly occurs on a speed-critical path, we feel that this tradeoff is a good one. The slight slowdown for small CREATE INDEX statements might be recovered in a future release. ORDER BY and GROUP BY operations should now be faster for all cases, large and small.
The query planner has been enhanced to do a better job of handling the DISTINCT keyword on SELECT statements.
There has been a lot of work on the default VFSes. The unix VFS has been enhanced to include more overrideable system calls - a feature requested by Chromium to make it easier to build SQLite into a sandbox. The windows VFS has been enhanced to be more resistant to interference from anti-virus software.
Every version of SQLite is better tested than the previous, and 3.7.8 is no exception to this rule. Version 3.7.8 has been used internally by the SQLite team for mission critical functions and has performed flawlessly. And, of course, it passes our rigorous testing procedures with no problems detected. Version 3.7.8 is recommended for all new development.
2011-06-28 - Version 3.7.7.1
SQLite version 3.7.7.1 adds a one-line bug fix to 3.7.7 to fix a problem causing PRAGMA case_sensitive_like statements compiled using the legacy sqlite3_prepare() interface to fail with an SQLITE_SCHEMA error. Because sqlite3_exec() uses sqlite3_prepare() internally, the problem also affects sqlite3_exec().
Upgrading from 3.7.7 is only required for applications that use "PRAGMA case_sensitive_like" and the sqlite3_prepare() (or sqlite3_exec()) interface.
2011-06-24 - Version 3.7.7
SQLite version 3.7.7 is a regularly scheduled bi-monthly maintenance release. Upgrading from version 3.7.6.3 is optional. Upgrading from all prior releases is recommended.
This release adds support for naming database files using URI filenames. URI filenames are disabled by default (for backwards compatibility) but applications are encouraged to enable them since incompatibilities are likely to be exceedingly rare and the feature is useful. See the URI filename documentation for details.
Most of the other enhancements in this release involve virtual tables. The virtual table interface has been enhanced to support SAVEPOINT and ON CONFLICT clause processing, and the built-in RTREE and FTS3/FTS4 have been augmented to take advantage of the new capability. This means, for example, that it is now possible to use the REPLACE command on FTS3/FTS4 and RTREE tables.
The FTS4 full-text index extension has been enhanced to support the FTS4 prefix option and the FTS4 order option. These two enhancements are provided in support of search-as-you-type interfaces where search results begin to appear after the first keystroke in the "search" box and are refined with each subsequent keystroke. The way this is done is to do a separate full-text search after each key stroke, and add the "*" wildcard at the end of the word currently being typed. So, for example, if the text typed so far is "fast da" and the next character typed is "t", then the application does a full-text search of the pattern "fast dat*" and displays the results. Such capability has always existed. What is new is that the FTS4 prefix option allows the search to be very fast (a matter of milliseconds) even for difficult cases such as "t*" or "th*".
There has been a fair amount of work done on the FTS4 module for this release. But the core SQLite code has changed little and the previous release has not given any problems, so we expect this to be a very stable release.
2011-05-19 - Version 3.7.6.3
SQLite version 3.7.6.3 is a patch release that fixes a single bug associated with WAL mode. The bug has been in SQLite ever since WAL was added, but the problem is very obscure and so nobody has noticed before now. Nevertheless, all users are encouraged to upgrade to version 3.7.6.3 or later.
The bug is this: If the cache_size is set very small (less than 10) and SQLite comes under memory pressure and if a multi-statement transaction is started in which the last statement prior to COMMIT is a SELECT statement and if a checkpoint occurs right after the transaction commit, then it might happen that the transaction will be silently rolled back instead of being committed.
The default setting for cache_size is 2000. So in most situations, this bug will never appear. But sometimes programmers set cache_size to very small values on gadgets and other low-memory devices in order to save memory space. Such applications are vulnerable. Note that this bug does not cause database corruption. It is as if ROLLBACK were being run instead of COMMIT in some cases.
Bug Details
Transactions commit in WAL mode by adding a record onto the end of the WAL (the write-ahead log) that contains a "commit" flag. So to commit a transaction, SQLite takes all the pages that have changed during that transaction, appends them to the WAL, and sets the commit flag on the last page. Now, if SQLite comes under memory pressure, it might try to free up memory space by writing changed pages to the WAL prior to the commit. We call this "spilling" the cache to WAL. There is nothing wrong with spilling cache to WAL. But if the memory pressure is severe, it might be that by the time COMMIT is run, all changed pages for the transaction have already been spilled to WAL and there are no pages left to be written to WAL. And with no unwritten pages, there was nothing to put the commit flag on. And without a commit flag, the transaction would end up being rolled back.
The fix to this problem was that if all changed pages has already been written to the WAL when the commit was started, then page 1 of the database will be written to the WAL again, so that there will always be a page available on which to set the commit flag.
2011-04-17 - Version 3.7.6.2
SQLite version 3.7.6.2 adds a one-line bug fix to 3.7.6.1 that enables pthreads to work correctly on NetBSD. The problem was a faulty function signature for the open system call. The problem does not appear to have any adverse impact on any system other than NetBSD.
Upgrading from version 3.7.6.1 is only needed on NetBSD.
2011-04-13 - Version 3.7.6.1
SQLite version 3.7.6.1 fixes a single bug in 3.7.6 that can cause a segfault if SQLITE_FCNTL_SIZE_HINT is used on a unix build that has SQLITE_ENABLE_LOCKING_MODE set to 0 and is compiled with HAVE_POSIX_FALLOCATE.
Upgrading from 3.7.6 is only needed for users effected by the configuration-specific bug described above. There are no other changes to the code.
2011-04-12 - Version 3.7.6
SQLite version 3.7.6 is a regularly scheduled bi-monthly maintenance release of SQLite. Upgrading from version 3.7.5 is optional. Upgrading releases prior to 3.7.5 is recommended.
2011-02-01 - Version 3.7.5
SQLite version 3.7.5 is a regularly scheduled bi-monthly maintenance release of SQLite. Due to the discovery and fix of an obscure bug that could cause database corruption, upgrading from all prior releases of SQLite is recommended. This bug was found during code review and has not been observed in the wild.
This release adds new opcodes for the sqlite3_db_status() interface that allow more precise measurement of how the lookaside memory allocator is performing, which can be useful for tuning in applications with very tight memory constraints.
The sqlite3_vsnprintf() interface was added. This routine is simply a varargs version of the long-standing sqlite3_snprintf() interface.
The output from sqlite3_trace() interface has been enhanced to work better (and faster) in systems that use recursive extensions such as FTS3 or RTREE.
Testing with Valgrind shows that this release of SQLite is about 1% or 2% faster than the previous release for most operations.
A fork of the popular ADO.NET adaptor for SQLite known as System.Data.SQLite is now available on http://System.Data.SQLite.org/. The originator of System.Data.SQLite, Robert Simpson, is aware of this fork, has expressed his approval, and has commit privileges on the new Fossil repository. The SQLite development team intends to maintain System.Data.SQLite moving forward.
2010-12-08 - Version 3.7.4
SQLite version 3.7.4 is a regularly scheduled bi-monthly maintenance release of SQLite. Upgrading from version 3.7.2 and version 3.7.3 is optional. Upgrading from all other SQLite releases is recommended.
This release features full-text search enhancements. The older FTS3 virtual table is still fully supported, and should also run faster. In addition, the new FTS4 virtual table is added. FTS4 follows the same syntax as FTS3 but holds additional metadata which facilitates some performance improvements and more advanced matchinfo() output. Look for further full-text search enhancements in subsequent releases.
Also in this release, the EXPLAIN QUERY PLAN output has been enhanced and new documentation is provided so that application developers can more easily understand how SQLite is performing their queries.
Thanks to an account from the folks at http://www.devio.us/, OpenBSD has been added to the list of platforms upon which we test SQLite prior to every release. That list of platforms now includes:
Linux x86 & x86_64
MacOS 10.5 & 10.6
MacOS 10.2 PowerPC
WinXP and Win7
Android 2.2
OpenBSD 4.7
The previous release of SQLite (version 3.7.3) has proven to be very robust. The only serious issue discovered was ticket 80ba201079 that describes an incorrect query result that can occur under very unusual circumstances. The ticket description contains details of the problem. Suffice it to say here that the problem is very obscure and is unlikely to effect most applications and so upgrading is optional. The problem is fixed, of course, in this release.
2010-October-08 - Version 3.7.3
SQLite version 3.7.3 is a regularly scheduled bi-monthly maintenance release of SQLite. Upgrading from version 3.7.2 is optional. Upgrading from all other releases is recommended.
This release adds two new interfaces (really just variations on existing interfaces). The sqlite3_create_function_v2() interface adds a destructor for the application-data pointer. The new sqlite3_soft_heap_limit64() interface allows the soft heap limit to be set to a value greater than 231.
The RTREE extension has been enhanced with the ability to have an application-defined query region. This might be used, for example, to locate all objects within the field of view of a camera.
The 3.7.3 release also includes some performance enhancements, including query planner improvements, documentation updates, and fixes to some very obscure bugs.
2010-August-24 - Version 3.7.2
SQLite version 3.7.2 fixes a long-standing bug that can cause the database free-page list to go corrupt if incremental_vacuum is used multiple times to partially reduce the size of a database file that contains many hundreds of unused database pages. The original bug reports together with links to the patch that fixes it can be seen here.
This bug has been in the code for at least a year and possibly longer. The bug has nothing to do with the versions 3.7.1 or 3.7.0 or any other recent release. The fact that the bug was discovered (and fixed) within hours of the 3.7.1 release is purely a coincidence.
The bug is impossible to hit without using incremental_vacuum and is very difficult to hit even with incremental_vacuum. And the kind of corruption that the bug causes can usually be fixed simply by running VACUUM. Nevertheless, because the bug can result in database corruption, it is recommended that all SQLite users upgrade to version 3.7.2 or later.
2010-August-23 - Version 3.7.1
SQLite version 3.7.1 is a stabilization release for the 3.7.x series. Other than the filesize-in-header bug that was fixed in version 3.7.0.1, no major problems have been seen in 3.7.0. Some minor corner-case performance regressions have been fixed. A typo in the OS/2 interface has been repaired.
A biggest part of the 3.7.1 release is a cleanup and refactoring of the pager module within SQLite. This refactoring should have no application-visible effects. The purpose was to reorganize the code in ways that make it easier to prove correctness.
The 3.7.1 release adds new experimental methods for obtained more detailed memory usage information and for controlling database file fragmentation. And the query planner now does a better job of optimizing the LIKE and GLOB operators.
This release increases the maximum size of database pages from 32KiB to 64KiB. A database with 64KiB pages will not be readable or writable by older versions of SQLite. Note that further increases in page size are not feasible since the file format uses 16-bit offsets to structures within each page.
2010-August-04 - Version 3.7.0.1
SQLite version 3.7.0.1 is a patch release to fix a bug in the new filesize-in-header feature of the SQLite file format that could cause database corruption if the same database file is written alternately with version 3.7.0 and version 3.6.23.1 or earlier. A performance regression was also fixed in this release.
2010-07-22 - Version 3.7.0
SQLite version 3.7.0 is a major release of SQLite that features a new transaction control mechanism using a write-ahead log or WAL. The traditional rollback-journal is still used as the default so there should be no visible change for legacy programs. But newer programs can take advantage of improved performance and concurrency by enabling the WAL journaling mode.
SQLite version 3.7.0 also contains some query planner enhancements and a few obscure bug fixes, but the only really big change is the addition of WAL mode.
2010-03-30 - Version 3.6.23.1
SQLite version 3.6.23.1 is a patch release to fix a bug in the offsets() function of FTS3 at the request of the Mozilla.
2010-03-09 - Version 3.6.23
SQLite version 3.6.23 is a regular bimonthly release of SQLite. Upgrading from the prior release is purely optional.
This release contains new pragmas: the secure_delete pragma, and the compile_options pragma. There are a new SQL functions: sqlite_compileoption_used() and sqlite_compileoption_get(). New C/C++ interfaces: sqlite3_compileoption_used(), sqlite3_compileoption_get(), SQLITE_CONFIG_LOG, and sqlite3_log().
This release also includes several minor bug fixes and performance improvements. Support for SQLITE_OMIT_FLOATING_POINT is enhanced. There are on-going improvements to FTS3.
The ".genfkey" command in the Command Line Interface has been removed. SQLite has supported standard SQL foreign key constraints since version 3.6.19 and so the ".genfkey" command was seen as an anachronism.
2010-01-06 - Version 3.6.22
SQLite version 3.6.22 is a bug-fix release. Two bugs have been fixed that might cause incorrect query results.
Ticket 31338dca7e describes a problem with queries that have a WHERE clause of the form (x AND y) OR z where x and z come from one table of a join and y comes from a different table.
Ticket eb5548a849 describes a problem where the use of the CAST operator in the WHERE clause can lead to incorrect results if the column being cast to a new datatype is also used in the same WHERE clause without being cast.
Both bugs are obscure, but because they could arise in an application after deployment, it is recommended that all applications upgrade SQLite to version 3.6.22.
This release also includes other minor bug fixes and performance enhancements, especially in the FTS3 extension.
2009-12-07 - Version 3.6.21
SQLite version 3.6.21 focuses on performance optimization. For a certain set of traces, this version uses 12% fewer CPU instructions than the previous release (as measured by Valgrind). In addition, the FTS3 extension has been through an extensive cleanup and rework and the sqlite3_trace() interface has been modified to insert bound parameter values into its output.
2009-11-04 - Version 3.6.20
SQLite version 3.6.20 is a general maintenance release. The query planner has been enhanced to work better with bound parameters in LIKE and GLOB operators and in range constraints and various minor bugs have been fixed. Upgrading from 3.6.19 is optional.
2009-10-14 - Version 3.6.19
SQLite version 3.6.19 adds native support for foreign key constraints, including deferred constraints and cascading deletes. Enforcement of foreign keys is disabled by default for backwards compatibility and must be turned on using the foreign_keys pragma.
Version 3.6.19 also adds support for the IS and IS NOT operators. Formerly, SQLite (as most other SQL database engines) supported IS NULL and IS NOT NULL. The IS and IS NOT operators are generalizations that allow the right-hand side to be an arbitrary expression. IS and IS NOT work the same as == (equals) and != (not equals) except that with IS and IS NOT the NULL values compare equal to one another.
2009-09-11 - Version 3.6.18
Beginning with this release, the SQLite source code is tracked and managed using the Fossil distributed configuration management system. SQLite was previously versioned using CVS. The entire CVS history has been imported into Fossil. The older CVS repository remains on the website but is read-only.
There are two major enhancements in SQLite version 3.6.18. The first is a series or refinements to the query planner that help SQLite to choose better plans for joins where in the past it was selecting suboptimal query plans. The SQLITE_ENABLE_STAT2 compile-time option has been added to cause SQLite to collect histogram data on indices when the ANALYZE command is run. The use of histograms improve the query planning performance even more.
The second major enhancement is that SQLite now support recursive triggers. The older non-recursive behavior of triggers is still the default behavior. Recursive triggers are activated using the recursive_triggers pragma. In addition to allowing triggers to call themselves (either directly or indirectly) the new capability also fires DELETE triggers on rows that are removed from a table as a result of REPLACE conflict resolution processing.
Non-recursive triggers are still the default behavior since this is least likely to cause problems for existing applications. However, we anticipate that triggers will become recursive by default beginning with release 3.7.0. At that point, applications that want to continue using the older non-recursive trigger behavior will need to use the recursive_triggers pragma to disable recursive triggers.
This version of SQLite also contains bug fixes, though none of the bugs are serious and all are obscure, so upgrading is optional.
The SQLite core continues to have 100% branch test coverage and so despite the many changes in this release, the developers believe that this version of SQLite is stable and ready for production use.
2009-08-10 - Version 3.6.17
This is a monthly maintenance release with a focus of bug fixes, performance improvements, and increased test coverage. This is the first release of SQLite since 100% branch test coverage was achieved on the SQLite core.
In addition, a new interface sqlite3_strnicmp() is provided for the convenience of extension writers.
None of the bugs fixed in this release are serious. All bugs are obscure. Upgrading is optional.
2009-07-25 - 100% Branch Test Coverage
A subset of the TH3 test suite was measured by gcov to provide 100% branch test coverage over the SQLite core (exclusive of the VFS backend and of extensions such as FTS3 and RTREE) when compiled for SuSE 10.1 Linux on x86. The SQLite developers pledge to maintain branch test coverage at 100% in all future releases. Ongoing work will strive for 100% branch test coverage on the operating-system backends and extensions as well.
2009-06-27 - Version 3.6.16
SQLite version 3.6.16 is another general maintenance release containing performance and robustness enhancements. A single notable bug was fixed (ticket #3929). This bug cause cause INSERT or UPDATE statements to fail on indexed tables that have AFTER triggers that modify the same table and index.
2009-06-15 - Version 3.6.15
SQLite version 3.6.15 is a general maintenance release containing performance and robustness enhancements and fixes for various obscure bugs.
2009-05-25 - Version 3.6.14.2
SQLite version 3.6.14.2 fixes an obscure bug in the code generator (ticket #3879) section of SQLite which can potentially cause incorrect query results. The changes from the prior release consist of only this one bug fix, check-in [6676] and a change to the version number text.
The bug was introduced in version 3.6.14. It is recommended that users of version 3.6.14 and 3.6.14.1 upgrade to this release. Applications are unlikely to hit this bug, but since it is difficult to predict which applications might hit it and which might not, we recommend that all users of 3.6.14 and 3.5.14.1 upgrade to this release.
2009-05-19 - Version 3.6.14.1
SQLite version 3.6.14.1 is a patch release to version 3.6.14 with minimal changes that fixes three bugs. Upgrading is only necessary for users who are impacted by one or more of those bugs.
2009-05-07 - Version 3.6.14
SQLite version 3.6.14 provides new performance enhancements in the btree and pager layers and in the query optimizer. Certain workloads can be as much as twice as fast as the previous release, though 10% faster is a more typical result.
Queries against virtual tables that contain OR and IN operators in the WHERE clause are now able to use indexing.
A new optional asynchronous I/O backend is available for unix and windows. The asynchronous backend gives the illusion of faster response time by pushing slow write operations into a background thread. The tradeoff for faster response time is that more memory is required (to hold the content of the pending writes) and if a power failure or program crash occurs, some transactions that appeared to have committed might end up being rolled back upon restart.
This release also contains many minor bug fixes, documentation enhancements, new test cases, and cleanups and simplifications to the source code.
There is no compelling reason to upgrade from versions 3.6.12 or 3.6.13 if those prior versions are working. Though many users may benefit from the improved performance.
Same trick as with atf/test-programs: provide hand-generated Atffile and
Kyuafile files so that the helpers that we build as test programs do not
end up in them.
Reported by gson@.
Yes, attempting yet another fix at this so that the version number that
gets recorded in the pkgconfig files and inside atf-version really matches
the latest imported version. Should resolve issues where the built files
get stuck with an older version number during update builds.
This time, I'm trying the same approach I applied in the FreeBSD source
tree, which has been working fine so far across various release imports.
They are not intended to be run neither by atf-run nor Kyua, and doing so
results in test failures. The easiest way to do this for now is to just
ship custom Atffile and Kyuafile files. (This broke because with atf-0.20
we started using the auto-generated versions of these, and due to the way
bsd.test.mk works, these registered the helpers as well.)
Problem reported by martin@.
The build would break when we do not use MAKEOBJDIR* but do use OBJMACHINE.
Problem found by B Harder and fix based on patch from NONAKA Kimihiro as
posted on current-users.
- Move the majority of the common build definitions to the top-level
Makefile.inc and ensure this gets included everywhere.
- Move the bconfig.h file to the top-level directory.
- Add a statically-generated defs.h file instead of creating one
during the build. Easier to understand and less chances for things
to go wrong.
- Make sure all files using ATF_VERSION have the right dependency to
trigger a rebuild when the value changes.
- Clean up stale -I flags.
This is all mostly for simplicity reasons and to reduce the cognitive
load required to understand the build of the atf and kyua-* packages.
I have tested this with both MKKYUA=no/yes and non-clean/clean builds
so hopefully I got the details right. But if not, let me know please.
Because we now own the 'tools' subdirectory in the tree, we can yank some
of the upstream autoconf-related complexity. Start doing so by removing
defs.hpp and using the real compiler attributes where necessary.
Experimental version released on February 7th, 2014.
This is the first release without the code for the deprecated tools. If
you require such code, please fetch a copy of the 0.19 release and extract
the 'tools' directory for your own consumption.
* Removed the deprecated tools. This includes atf-config, atf-report,
atf-run and atf-version.
The main change here is that the atf-config, atf-report, atf-run and
atf-version tools no longer depend on libatf-c nor libatf-c++. Instead,
they depend on an internal libtools.a that contains code specifically
for these tools and nothing else, making them self-contained.
Changes in version 0.19
***********************
Experimental version released on February 7th, 2014.
This is the last release to bundle the code for the deprecated tools.
The next release will drop their code and will stop worrying about
backwards compatibility between the ATF libraries and what the old tools
may or may not support.
If you still require the old tools for some reason, grab a copy of the
'tools' directory now. The code in this directory is standalone and
does not depend on any internal details of atf-c++ any longer.
* Various fixes and improvements to support running as part of the FreeBSD
test suite.
* Project hosting moved from Google Code (as a subproject of Kyua) to
GitHub (as a first-class project). The main reason for the change is
the suppression of binary downloads in Google Code on Jan 15th, 2014.
See https://github.com/jmmv/atf/
* Removed builtin help from atf-sh(1) and atf-check(1) for simplicity
reasons. In other words, their -h option is gone.
* Moved the code of the deprecated tools into a 'tools' directory and
completely decoupled their code from the internals of atf-c++. The
reason for this is to painlessly allow a third-party to maintain a
copy of these tools after we delete them because upcoming changes to
atf-c++ would break the stale tools.
Changes in version 0.18
***********************
Experimental version released on November 16th, 2013.
* Issue 45: Added require.memory support in atf-run for FreeBSD.
* Fixed an issue with the handling of cin with libc++.
* Issue 64: Fixed various mandoc formatting warnings.
* NetBSD PR bin/48284: Made atf-check flush its progress message to
stdout so that an interrupted test case always shows the last message
being executed.
* NetBSD PR bin/48285: Fixed atf_check examples in atf-sh-api(3).
- Delete some unnecessary files.
- Print out both the list of deleted files and added files in the import.
Useful when adjusting file lists and the reachover Makefiles.
outside */dist/* subdirectories.
When USE_PIGZGZIP=yes, bsd.own.mk sets TOOL_GZIP=${TOOL_PIGZ},
so there's no need to test USE_PIGZGZIP in these Makefiles.
