|
out of date - it was based on a.out OBJECT_FMT, and added entries in the
generated PLISTs to reflect the symlinks that ELF packages uses. It also
tried to be clever, and removed and recreated any symbolic links that were
created, which has resulted in some fun, especially with packages which
use dlopen(3) to load modules. Some recent changes to our ld.so to bring
it more into line with other Operating Systems also exposed some cracks.
+ Modify bsd.pkg.mk and its shared object handling, so that PLISTs now contain
the ELF symlinks.
+ Don't mess about with file system entries when handling shared objects in
bsd.pkg.mk, since it's likely that libtool and the BSD *.mk processing will
have got it right, and have a much better idea than we do.
+ Modify PLISTs to contain "ELF symlinks"
+ On a.out platforms, delete any "ELF symlinks" from the generated PLISTs
+ On ELF platforms, no extra processing needs to be done in bsd.pkg.mk
+ Modify print-PLIST target in bsd.pkg.mk to add dummy symlink entries on
a.out platforms
+ Update the documentation in Packages.txt
With many thanks to Thomas Klausner for keeping me honest with this.
|
|
--- 9.1.0b2 released ---
641. [bug] $GENERATE caused a uninitialized link to be used.
[RT #595]
640. [bug] Memory leak in error path could cause
"mpctx->allocated == 0" failure. [RT #584]
639. [bug] Reading entropy from the keyboard would sometimes fail.
[RT #591]
638. [port] lib/isc/random.c needed to explicitly include
time.h explicitly to get a prototype for time() when
pthreads was not being used. [RT #592]
637. [port] Use isc_u?int64_t instead of (unsigned) long long in
lib/isc/print.c. Also allow lib/isc/print.c to
be compiled even if the platform does not need it.
[RT #592]
636. [port] Shut up MSVC++ about a possible loss of precision
in the ISC__BUFFER_PUTUINT*() macros. [RT #592]
635. [bug] Reloading a server with a configured blackhole list
would cause an assertion. [RT #590]
634. [bug] A log file will completely stop being written when
it reaches the maximum size in all cases, not just
when versioning is also enabled. [RT #570]
633. [port] Cope with rlim_t missing on BSD/OS systems. [RT #575]
632. [bug] The index array of the journal file was
corrupted as it was written to disk.
631. [port] Build without thread support on systems without
pthreads.
630. [bug] Locking failure in zone code. [RT #582]
629. [bug] 9.1.0b1 dereferenced a null pointer and crashed
when responding to a UDP IXFR request.
628. [bug] If the root hints contained only AAAA addresses,
named would be unable to perform resolution.
627. [bug] The EDNS0 blackhole detection code of changed 324
waited for three retransmissions to each server,
which takes much too long when a domain has many
name servers and all of them drop EDNS0 queries.
Now we retry without EDNS0 after three consecutive
timeouts, even if they are all from different
servers. [RT #143]
626. [bug] The lightweight resolver daemon no longer crashes
when asked for a SIG rrset. [RT #558]
625. [func] Zones now inherit their class from the enclosing view.
624. [bug] The zone object could get timer events after it had
been destroyed, causing a server crash. [RT #571]
623. [func] Added "named-checkconf" and "named-checkzone" program
for syntax checking named.conf files and zone files,
respectively.
622. [bug] A canceled request could be destroyed before
dns_request_destroy() was called. [RT #562]
621. [port] Disable IPv6 at runtime if IPv6 sockets are unusable.
This mostly affects Red Hat Linux 7.0, which has
conflicts between libc and the kernel.
620. [bug] dns_master_load*inc() now require 'task' and 'load'
to be non-null. Also 'done' will not be called if
dns_master_load*inc() fails immediately. [RT #565]
618. [bug] Queries to a signed zone could sometimes cause
an assertion failure.
617. [bug] When using dynamic update to add a new RR to an
existing RRset with a different TTL, the journal
entries generated from the update did not include
explicit deletions and re-additions of the existing
RRs to update their TTL to the new value.
616. [func] dnssec-signzone -t output now includes performance
statistics.
615. [bug] dnssec-signzone did not like child keysets signed
by multiple keys.
614. [bug] Checks for uninitialized link fields were prone
to false positives, causing assertion failures.
The checks are now disabled by default and may
be re-enabled by defining ISC_LIST_CHECKINIT.
613. [bug] "rndc reload zone" now reloads primary zones.
It previously only updated slave and stub zones,
if an SOA query indicated an out of date serial.
612. [cleanup] Shutup a ridiculously noisy HP-UX compiler that
complains relentlessly about how its treatment
of 'const' has changed as well as how casting
sometimes tightens alignment constraints.
611. [func] allow-notify can be used to permit processing of
notify messages from hosts other than a slave's
masters.
610. [func] rndc dumpdb is now supported.
609. [bug] getrrsetbyname() would crash lwresd if the server
found more SIGs than answers. [RT #554]
608. [func] dnssec-signzone now adds a comment to the zone
with the time the file was signed.
607. [bug] nsupdate would fail if it encountered a CNAME or
DNAME in a response to an SOA query. [RT #515]
606. [bug] Compiling with --disable-threads failed due
to isc_thread_self() being incorrectly defined
as an integer rather than a function.
605. [func] New function isc_lex_getlasttokentext().
604. [bug] The named.conf parser could print incorrect line
numbers when long comments were present.
603. [bug] Make dig handle multiple types or classes on the same
query more correctly.
602. [func] Cope automatically with UnixWare's broken
IN6_IS_ADDR_* macros. [RT #539]
601. [func] Return a non-zero exit code if an update fails
in nsupdate.
600. [bug] Reverse lookups sometimes failed in dig, etc...
599. [func] Added four new functions to the libisc log API to
support i18n messages. isc_log_iwrite(),
isc_log_ivwrite(), isc_log_iwrite1() and
isc_log_ivwrite1() were added.
598. [bug] An update-policy statement would cause the server
to assert while loading. [RT #536]
597. [func] dnssec-signzone is now multithreaded.
596. [bug] DNS_RDATASLAB_FORCE and DNS_RDATASLAB_EXACT are
not mutually exclusive.
595. [port] On Linux 2.2, socket() returns EINVAL when it
should return EAFNOSUPPORT. Work around this.
[RT #531]
594. [func] sdb drivers are now assumed to not be thread-safe
unless the DNS_SDBFLAG_THREADSAFE flag is supplied.
593. [bug] If a secure zone was missing all its NXTs and
a dynamic update was attempted, the server entered
an infinite loop.
592. [bug] The sig-validity-interval option now specifies a
number of days, not seconds. This matches the
documentation. [RT #529]
|
