"PostMark: A New File System Benchmark" -- Existing file system benchmarks
are deficient in portraying performance in the ephemeral small-file regime
used by Internet software, especially: electronic mail, netnews, and
web-based commerce.
Allow ${PREFIX} setting. (ports/12213)
Clean up makefile a smidge.
PR: ports/11581 and ports/12213
Submitted by: Martin Kammerhofer dada@sbox.tu-graz.ac.at
and Nick Hibma nick.hibma@jrc.it
FWIW, checkout of these things took 5+hrs, staying on the local
.freebsd.org net w/o hitting the 'net at all.
As promised,
$ time cvs ci
real 67m51.701s
user 0m1.250s
sys 0m5.345s
distribution includes a high-performance client and server simulators.
The simulators create a stream of HTTP requests that can be routed through
a Web proxy. Studying proxy performance under various [stress] conditions is
essential for performance tuning, evaluation of new algorithms, analysis of
hardware configurations, and comparing available proxy products.
PR: ports/10375
Submitted By: Dima Sivachenko <dima@Chg.RU>
bsd.port.mk rev. 1.304 for details on the change.
The fix here is one of the following.
(1) Define USE_BZIP2 instead of BUILD_DEPENDS on bzip2 and redefining
EXTRACT_* commands.
(2) Change ${EXTRACT_CMD} to ${TAR} when the command is obviously
calling the "tar" command (i.e., arguments like "-xzf" are spelled
out).
(3) If ${EXTRACT_CMD} is called directly with ${EXTRACT_BEFORE_ARGS},
add ${EXTRACT_AFTER_ARGS} to the command line as well.
(4) If any of EXTRACT_CMD, EXTRACT_BEFORE_ARGS or EXTRACT_AFTER_ARGS
is set, define the other two too.
This program is an up to date version of the ttcp program.
It uses inetd (or simulates its behaviour) to start off the remote
side program which will send/receive data. Both sides measure the time
and number of bytes transfered. The local side will print the measures.
The format of the output can be specified on the commandline.
PR: ports/8546
I eliminated DISTFILES and PKGNAME by just using EXTRACT_SUFX.
On a funnier note, the MASTER_SITE didn't like us sending our e-mail
address as "president@whitehouse.gov" to get around it's "no-root" login
policy. So I had to change this to "portsuser@FreeBSD.org".
530- SIR, how'bout using a NSA machine instead?
530 Goodbye.
Don't ask me how this differs from bytebench, but it seems to be
different enough to warrant a seperate port.
PR: 5330
Submitted by: Andrey Zakhvatov <andy@icc.surw.chel.su>
* PKGNAME needs a version number
* DISTNAME useless
* use INSTALL_* macros.
* "mkdir -p" --> ${MKDIR}
* changed "BTW" to something non-native english speakers may understand.
It displays a part of a motor. The faster your X server is, the more
rpm you get ;-) Guess how many rpm's I got, when I started with X11
years ago .... 32 rpm with an ET4000 VGA card on a 386 4MB running
Interactive Unix ;-))
PLISTs.
Note: I know that this is going to break some symlinks and/or .so
includes, I will back some of these out as I run into these during
package building.
all the COMMENTs! No package names, no version numbers, no "this is
absolutix-3.1.2" type comments that have zero information contents.
Now, without any bad examples to follow, nobody has an excuse to import
a port with those kind of comments. :)
Phew! 238 ports modified!
- Temporary files are written into $TMPDIR (default /var/tmp)
(some 18MB of free space are required for the file system test).
- Results are stored into $RESULTDIR (default /tmp).
prevent the tests from running.
Do not try to compile the test programs, since the sources are not
being copied to the lib/bytebench directory by "make install".
(See: http://www.silkroad.com/bass/linux/bm.html for results ...)
This is another benchmark that tries to give a single performance
number, but without giving much thought to proper weighting of the
performance numbers. So: Please handle with care ...
Yes, I know the package files are missing, I need to hack it a
little more so that it installs the binaries and all the data files
in some place that won't be blown away by a "make clean" and compare
the results...for now, you can type "make test" to run the test and
go into the work/lmbench-1.0/Results subdirectory to see what it
measured.
(1) Took out INSTALL_MANPAGES (not necessary anymore, porter should
set NO_INSTALL_MANPAGES for not calling "make install.man")
(2) Replaced most of DEPENDS with EXEC_DEPENDS and LIB_DEPENDS. These
are the entries I used:
EXEC_DEPENDS:
unzip:${PORTSDIR}/archivers/unzip
gmake:${PORTSDIR}/devel/gmake
wishx:${PORTSDIR}/lang/tclX
xli:${PORTSDIR}/graphics/xli
gs:${PORTSDIR}/print/ghostscript
gunshar:${PORTSDIR}/archivers/gshar+gunshar
hfs:${PORTSDIR}/utils/hfs
rman:${PORTSDIR}/utils/rman
LIB_DEPENDS:
tiff\\.3\\.:${PORTSDIR}/graphics/tiff
jpeg\\.5\\.:${PORTSDIR}/graphics/jpeg
Xpm\\.4\\.:${PORTSDIR}/graphics/xpm
tcl\\.7\\.:${PORTSDIR}/lang/tcl
tk\\.3\\.:${PORTSDIR}/x11/tk
xview\\.1\\.:${PORTSDIR}/x11/xview-lib
Xaw3d\\.:${PORTSDIR}/x11/Xaw3d
mpeg\\.1\\.:${PORTSDIR}/graphics/mpeg-lib
xview\\.3\\.:${PORTSDIR}/x11/xview-lib
BLT\\.1\\.:${PORTSDIR}/x11/blt
There are still some dependencies I can't figure out what exactly
is needed. If your port still has DEPENDS in it, please check it out!
1. Make MASTER_SITES entries contain a trailing /. Garrett says that
assuming a '/' seperator between entries in MASTER_SITES and entries
in DISTFILES is a dangerous assumption. This will also be taken
out of bsd.port.mk.
2. General clean-up. Some of these Makefiles were a little grim.
Make sure they're all in sync with the sometimes rapidly changing
bsd.port.mk! :)
3. Some small configuration tweaks to keep them compiling under the
most recent 2.0 - some things seem to have broken along our road
to release.
