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.
the owner of all installed files is a non-root user. This change
affects most packages that require special users or groups by making
them use the specified unprivileged user and group instead.
(1) Add two new variables PKG_GROUPS_VARS and PKG_USERS_VARS to
unprivileged.mk. These two variables are lists of other bmake
variables that define package-specific users and groups. Packages
that have user-settable variables for users and groups, e.g. apache
and APACHE_{USER,GROUP}, courier-mta and COURIER_{USER,GROUP},
etc., should list these variables in PKG_USERS_VARS and PKG_GROUPS_VARS
so that unprivileged.mk can know to set them to ${UNPRIVILEGED_USER}
and ${UNPRIVILEGED_GROUP}.
(2) Modify packages to use PKG_GROUPS_VARS and PKG_USERS_VARS.
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).
around at either build-time or at run-time is:
USE_TOOLS+= perl # build-time
USE_TOOLS+= perl:run # run-time
Also remove some places where perl5/buildlink3.mk was being included
by a package Makefile, but all that the package wanted was the Perl
executable.
And always is defined as share/examples/rc.d
which was the default before.
This rc.d scripts are not automatically added to PLISTs now also.
So add to each corresponding PLIST as required.
This was discussed on tech-pkg in late January and late April.
Todo: remove the RCD_SCRIPTS_EXAMPLEDIR uses in MESSAGES and elsewhere
and remove the RCD_SCRIPTS_EXAMPLEDIR itself.
under share/examples/rc.d. The variable name already was named
RCD_SCRIPTS_EXAMPLEDIR.
This is from ideas from Greg Woods and others.
Also bumped PKGREVISION for all packages using RCD_SCRIPTS mechanism
(as requested by wiz).
The 3.17.3 release has no external api changes (but does fix a few api
implementation bugs so check the details below).
New minor features:
1) Add new option to spflooder. The -n option allows a fixed
"number of members" to be set. This then activates a multi-sender flow
control algorithm to allow flooding tests with several senders. All of
the processes need to join the group (i.e. they cannot be
-wo (write-only)) but not all processes have to send. This allows easy
testing with differing numbers of senders (just change how many
spflooders start with -ro and how many do not).
2) Modify monitor so it will run correctly on Windows. This was done by
modifying monitor.c so it can also build as a threaded program and
on windows will use multiple threads to send/recv updates and get
user input. (A POSIX thread option is also added).
3) Add new Alarm priority flag to print a line with no datestamp
(for multi-line output).
4) Add new Windows VC++.Net project files to daemon/win32_msvc_net.
The current project files in daemon/win32 remain as they can be used
in VC++ (version 6).
The list of bugfixes is:
1) Fix memory leak in Skiplist. Reported by Taj Khattra, patch by Theo
Schlossnagle.
2) spuser,spflooder, and spmonitor fixed to print correct name in help.
Patch by Daniel Rall.
3) Fix incorrect alarm printing where WARNING messages generated by
older Alarm() interface were not printed.
4) Fix a bug in the Windows build using VC++ of thread-safe libtspread.lib.
Bug and fix by Jacob Green.
5) Fix bug in libspread where if the groups array or message body passed
to SP_recv* was too small, the mess_type field returned would be truncated
and the sender field was not returned. They are both now returned
correctly. Bug report and partial fix provided by John Schultz.
6) Fix bug where SP_Join and SP_Leave do not report an error if a group name
is too long (instead they truncated it) Reported with fix by David Parker.
** Warning, this could break buggy applications who use long groups and
assume the name is truncated.
7) Cleanup compile warnings where E_queue() used with no-parameter
functions
(not all uses fixed) and fix incorrect use of signed int with strlen().
8) Fix few cases in flooder.c and user.c that did not use the defined
MAX_MESSLEN constant. Tested to verify that increasing
MAX_SCATTER_ELEMENTS in scatter.h and the MAX_MESSLEN defines in user.c
and flooder.c is sufficient to support arbitrarily large message sizes
with Spread. This is NOT recommended, but several people do it anyway :-)
9) Make E_delay() work on Win32.
10) Added check that a segment using localhost (i.e. 127.0.0.255) is NOT in
a spread.conf with any other segments using real IP addresses. This
configuration will not work correctly and is usually because the example
localhost configuration provided with Spread is added to. This check will
cause the daemon to exit immediately after parsing the config file.
11) Fix bug where if more then 22 daemons start at the same time, some will
crash or the membership will not complete correctly. This bug was
reported by several people including Jesse Noller.
12) Fix Java spread connection problem where several threaded connections
from same process cause unexpected connection failures. Reported by
Brian Moseley, idea of fix by Ryan Caudy, patch by Jonathan Stanton.
13) Fix EVS bug where AGREED messages may be delivered before a transitional
signal on some daemons and after it on others. Bug found and patch
created by Ryan Caudy.
14) Make SP_connect_timeout() calls non-blocking for the actual 'connect()'
call. This should fix the issue reported by Shlomi Yaakobovich where a
hung daemon causes new connections to also hang in connect. Also includes
slight cleanup of connect code path.
Ver 3.17.2:
--------------
1) Fix daemon quit when multiple interfaces are configured as "D" daemon
interfaces in the spread.conf file. Bug reported by Orit Wasserman.
2) Updated url for Java 'ant' build system. Patch by Daniel Rall.
3) Fix group_id bug that causes incorrect vs_sets. Patch by Ryan Caudy.
4) Fix spread.conf parser so it validates the machine names in segments
and forces them to be less then MAX_PROC_NAME. Patch by Mikhail Terekhov.
5) Minor fix to Mac OS X compilation so library softlinks do not fail the
second time make is run.
6) Alarm() changes to support priority levels on each Alarm() call.
7) Fix crash by improving packet accounting when a client connected to a
singleton daemon sends a large broadcast. Reported by David Shaw.
8) Fix bus errors on Sparc & Alpha for message buffer integer assignment.
Reported by Greg Shebert; tested and patched Mikhail Terekhov.
9) Verify daemon names in spread.conf are unique. If non-unique names are
provided in spread.conf, configuration will be rejected and daemon will
not start. Suggested by Tim Peters.
10) Zero buffer in c library before sending multicast.
Reported by Panagiotis Kougiouris.
11) Send fewer lookup probe messages when only a single segment is configured.
12) Remove extra token rotations when no messages are sent. Will decrease
network packet overhead.
13) Make mailbox and service in sp.h a typedef instead of a #define. Suggested
and patched by Steven Dake.
14) Fix small endianness error in sp.c where the mess_type field may not be
correctly converted for different endian platforms when the SP_*_recv calls
return a BUFFER_TOO_SHORT or GROUPS_TOO_SHORT error.
15) Change alarm tag for security prints from SEC to SECURITY because of conflict
with sys/time.h header.
16) Documentation fix to SP_receive man page to correct fields for self-leave
membership messages.
17) Update of email addresses in copyright statements and headers.
18) Windows binary libraries now built as libspread and libtspread like other
platforms.
the RCD_SCRIPTS rc.d script(s) to the PLIST.
This GENERATE_PLIST idea is part of Greg A. Woods'
PR #22954.
This helps when the RC_SCRIPTS are installed to
a different ${RCD_SCRIPTS_EXAMPLEDIR}. (Later,
the default RCD_SCRIPTS_EXAMPLEDIR will be changed
to be more clear that they are the examples.)
These patches also remove the etc/rc.d/ scripts from PLISTs
(of packages that use RCD_SCRIPTS). (This also removes
now unused references from openssh* makefiles. Note that
qmail package has not been changed yet.)
I have been doing automatic PLIST registration for RC_SCRIPTS
for over a year. Not all of these packages have been tested,
but many have been tested and used.
Somethings maybe to do:
- a few packages still manually install the rc.d scripts to
hard-coded etc/rc.d. These need to be fixed.
- maybe remove from mk/${OPSYS}.pkg.dist mtree specifications too.
"Package Makefiles should refer to PKG_SYSCONFBASEDIR instead of
PKG_SYSCONFBASE when they want PKG_SYSCONFDIR stripped of
PKG_SYSCONFSUBDIR. This makes PKG_SYSCONFBASE=/etc work with pkgviews by
installing all config files into /etc/packages/<pkg> instead of
occasionally putting some directly into /etc."
The NetBSD project now has the permission to download the archive from
the official Spread site and to mirror it, so remove the restrictions.
changes:
*) Fix memory corruption and crash with groups of large size.
*) Correct make install so it installs header files.
*) Fix syntax error in build.xml file for Java/Ant.
*) Cleanup prototypes to remove compiler warnings.
*) Fix parser to correctly recognize upper, lower, and mixed case command options.
*) During make install, remove old symlinks.
*) Change setgroups call to be more portable. (fixes MacOSX)
*) Change name of r and s to sprecv and spsend, and add as make targets.
They can be built by "make testprog" (not built by default).
*) Work on making long group names possible.
*) Increase listen backlog for accepting client connections.
*) Fix Win32 project files to have correct path to source files.
(note CVS was always ok, but 3.17.0 release had incorrect path)
*) Fix bug where large groups overflow Mess_buf in groups.c.
*) Fix memory corruption bug when a message header is received in
several separate packets in session.c. Thanks to Ryan Caudy for
many, many hours tracking this down.
*) Change order of build in Makefile so binaries are built before
documentation.
*) Fix Java bug where connection objects cannot be disconnected and
then reconnected, but must be created anew. They can now be reused.
*) Fix compile error on AIX for struct if_info.
*) Fix security issue with buffer checks in the C library.
*) Fix obscure off-by-one buffer error with the parser.
Spread is a group communication system that provides a number of
messaging services to applications. These services can then be used
by applications to make it easy to write full-featured reliable
groupware, distributed servers, client-server applications, chat
and news services, and many other networked enabled tools.
