if it already exists. Since the bootstrap JDK can be an already
installed instance of itself this creates a circular dependency which
breaks portmaster and possibly other port management tools.
. Actually create a valid BUILD_DEPENDS for the bootstrap JDK instead of
just skipping it.
. Make the default bootstrap JDK diablo-jdk16 rather than diablo-jdk15.
We should actually try and get this from bsd.java.mk somehow.
. Fix some portlint warnings:
. No quotes for RESTRICTED and NO_CDROM.
. Use variables for executables rather than the raw names.
. If DISTFILES are missing then print out the instructions to get them
nicely formatted and error out rather than trying to use IGNORE and
resetting ECHO_MSG.
. Fix some grammar errors in various messages printed out by the port.
confuses portmaster, which (understandably) expects that we didn't
fake up our bootstrap dependency.
Instead, only set up a BUILD_DEPENDS for a bootstrap if there isn't one
installed. A more complicated and arguably more correct patch would
always set up the BUILD_DEPENDS for the bootstrap JDK but would do so
correctly based on what it found to use.
This almost certainly needs to happen to the other jdk* ports, but I'm
going to await some positive feedback before doing so.
The affected ports are the ones with gettext as a run-dependency
according to ports/INDEX-7 (5007 of them) and the ones with USE_GETTEXT
in Makefile (29 of them).
PR: ports/124340
Submitted by: edwin@
Approved by: portmgr (pav)
- Remove USE_XLIB/USE_X_PREFIX/USE_XPM in favor of USE_XORG
- Remove X11BASE support in favor of LOCALBASE or PREFIX
- Use USE_LDCONFIG instead of INSTALLS_SHLIB
- Remove unneeded USE_GCC 3.4+
Thanks to all Helpers:
Dmitry Marakasov, Chess Griffin, beech@, dinoex, rafan, gahr,
ehaupt, nox, itetcu, flz, pav
PR: 116263
Tested on: pointyhat
Approved by: portmgr (pav)
This allows jconsole to show valid statistics for the following
OperatingSystem MBean statistics (some of these statistics are also shown
in the Summary and VM tab of jconsole):
TotalSwapSpaceSize, FreeSwapSpaceSize, FreePhysicalMemorySize and
TotalPhysicalMemorySize.
Note that MaxFileDescriptorCount has always been correct as far as I
can tell.
. Also, implement the function for ProcessCpuTime using a non-deprecated
function (getrusage(2) rather than times(3)).
These changes are restricted to 6.x and higher and have only been tested
on 6.x (where they appear to work correctly). The changes are not valid
for 4.x. Their validity on 5.x and 7.x is unknown (testing welcome).
The PR is not fully addressed by these changes since
CommittedVirtualMemorySize and OpenFileDescriptorCount remain bogus.
Suggestions on how to get these without using kvm(3) would be appreciated.
PR: 118735
passing -XX:+UseThreadPriorities.
. Remove the os_sleep hack which was used on FreeBSD to make sure lower
priority threads got time slices. Instead, just call pthread_yield().
On FreeBSD 7.x with libthr, this will still give lower priority threads
some time (with the above flag turned on), although such behaviour is
not guaranteed by POSIX. This boosts FreeBSD performance by 7-fold on
an 8 core system, putting it on a par with Solaris (benchmarks by kris@).
The Java standard and the JCK tests are somewhat contradictory on thread
priority being guaranteed to work, and in this case the performance
benefits appear to outweigh any possible side effects.
. Pick up DEFAULT_LD_LIBRARY_PATH from the build environment rather than
patching it into a file at build time. This simplifies the Makefile.
Submitted by: Kurt Miller <kurt@intricatesoftware.com>
appears to actually match the type of variable its being used with in the
code (although it seems that it may be even cleaner to just replace 'fastInt'
with 'short').
This also partially bandaids a build problem some are seeing since Hint.h
is a less common header file name than cpu.h.
