| Age | Commit message (Collapse) | Author | Files | Lines |
|---|
|
|
|
|
|
used as a drop-in replacement for 'autoconf' without unnecessarily updating
dependencies all over the place.
|
|
|
|
|
|
you would see something like this from "lintpkgsrc -i":
Version mismatch: 'libtool-base' 1.4.20010614nb8 vs 1.4.20010614UNDEFINEDnb8
|
|
- change to my email address
Changes since p5-Test-Simple-0.42
- Adrian Howard added TB->maybe_regex()
- Adding Mark Fowler's suggestion to make diag() return
false.
- TB->current_test() still not working when no tests were run via
TB itself. Fixed by Dave Rolsky.
|
|
- fmt on DESCR
Changes since p5-Expect-1.10 :
- removed soft_close() from DESTROY. Being overly nice to a doomed process
doesn't pay off. Old behaviour is available via
$Expect::Do_Soft_Close = 1;
+ added various aliases for functions starting with 'exp_'
+ added a hook for log_file: can be set to a code ref.
+ added various FAQ entries
+ added autoflush(1) to log_file
+ split 'new' and 'spawn' to be able to set slave pty params via stty
before actually spawning the program
+ added slave_pty()
+ added print_log_file(), send() now no longer prints to log file or
stdout.
+ added alarm to test.pl to avoid blocking on cygwin.
+ rearranged and changed tests to better suit the various systems
+ added rudimentary 'notransfer' option; global only, not on
per-pattern-basis; workaround available in FAQ
+ timeout handlers now also can exp_continue
+ added 'raw_pty' option, also setting master to raw if isatty()
+ added and corrected test for exit status; got rid of Test.pm
+ updated docs & FAQs; explained how terminal sizes and SIGWINCH
should be propagated
|
|
ChangeLog :
* Tty.pm, Pty.pm: v1.02; disable warning for non-existant die handler
* Makefile.PL: remove cpp, test-compile instead
* Tty.pm, Pty.pm: disable die handler when requiring Stty
|
|
|
|
now uses the standard way.
|
|
|
|
autom4te tool and speedupa.
|
|
In addition to changes in the package as distributed, we no longer install
the optional tasks from this package -- what tasks got included in the
optional task jarfile was highly dependent on what other packages the user
had installed. It is neither reasonable to depend on all of these packages
(many of them themselves built with ant!) in the ant package, nor reasonable
for two builds of the ant optional tasks to vary so widely.
In a few days, I will import a jakarta-ant-optional package, which depends
on a wide range of other packages, and builds all those optional tasks
currently supportable by pkgsrc.
Relevant changes since jakarta-ant 1.3 (the last pkgsrc version):
(For full changes, see ${LOCALBASE}/share/doc/java/ant/WHATSNEW)
* XmlLogger and <antstructure> now add an encoding declaration to the
XML files they generate.
* <fileset> has a new attribute "casesensitive" to make it match
filenames in a case insensitive way (if you set it to false) - by
default filesets remain case sensitive.
* <rmic> is now implemented using a factory. This makes extending
rmic to use a new compiler a lot easier but may break custom
versions of this task that rely on the old implementation.
* several Zip methods have changed their signature as we now use a Zip
package of our own that handles Unix permissions for directories.
Furthermore <zip> will now use the platform's default character
encoding for filenames - this is consistent with the command line
ZIP tools, but causes problems if you try to open them from within
Java and your filenames contain non US-ASCII characters. Use the new
encoding attribute of the task and set it to UTF8 to get the old
behavior.
* Ant will no longer use the canonical version of a path internally -
this may yield different results on filesystems that support
symbolic links.
* Pattern matching rules have changes slightly, the pattern foo*
doesn't match files contained in a directory named foo - use foo/*
instead.
* <fixcrlf> will not remove trailing whitespace at the end of lines anymore.
* Invalid manifest files will now cause build failures in the <jar> task.
* A delete task like
<delete includeEmptyFilesets="true">
<fileset dir="somedir" />
</delete>
will now remove "somedir" as well, unless there are still files left
in it (matched by the default excludes).
* The copy task will now fail if the file to be copied is not found.
* Ant properties defined in properties files now behave the same way as
properties defined in the build file. In particular the $ character needs
to be escaped in property values by doubling it to $$. So, to define a
property with the value $hello, you need to define it in a properties file
as
test.prop=$$hello
This was not the case in Ant 1.3
* <tar> will now add empty directories as well
* <touch> can now work on <fileset>s
* <uptodate> now supports a value attribute
* <fail> supports nested text
* <fixcrlf> won't override files that are already in the correct
format.
* <sql> now supports REM comments as well as // and --
* <jar> now has a nested <metainf> element following the same idea as
<war>'s <webinf>.
* <available> now has a "type" attribute you can use in conjunction
with the "file" attribute to specify whether the "file" you're
looking for is a file or a directory.
* Allow the <sql> Delimiter to be set in the so that Oracle stored procs may be
entered
* <execon> and <apply> can now optionally skip empty filesets.
* <javadoc> has a new useexternalfile attribute that makes it use a
temporary file for sourcefile and package names - helps to defeat
command line length limitations.
* Data types like <path> can now be defined inside of <target>s
* New filesonly attribute for <zip> and friends to suppress directory
entries.
* New update attribute for <zip> and friends - update an existing
archive instead of creating a new one.
* <apply> and <execon> have been merged into a single task.
* <available> has a new filepath attribute/nested element that allows
you top search for a file in a given path.
* <taskdef> can now define several tasks at once, reading the
name/classname pairs from a property file or resource.
* <unzip/unjar/unwar> and <untar> now have an overwrite attribute that
defaults to true. If set to false, files that are newer than the
files in the archive will not be replaced.
* <patternset> and <fileset> now support nested <in/excludesfile>
elements - using these you can have more than one in/excludes file
per <patternset>.
* <fixcrlf> can now with CR only line-ends and can use an arbitrary
between 2 and 80.
* <move> will now try to rename() files before copying them byte by
byte - only if filtering is of, of course.
* <ant> and <antcall> tasks now support a new attribute inheritAll. When set to
false, only user properties are passed through to the target Ant instance.
This includes properties set on the command line and properties explicitly
passed
* <javadoc> now skips off line links if the package list cannot be found.
* <java> task will set the Thread contextClassLoader under JDKs 1.2+ to the
classloader for the class being executed.
* Introduce the concept of a TaskContainer - a task or element which can contain
Ant Tasks.
* Add new tasks implementing the TaskContainer interface <parallel> and
<sequential> which allow parallel execution of tasks to be specified.
* <depend> task will now take into account dependencies on jar files and class
files from a given classpath.
* <jar> manifest entries may now be specified in the build file either
completely or to be merged with a manifest file.
* <tstamp> task custom formats now support locales.
* Introduced the concept of <filtersets> to allow for more control in which
filters get applied in a <copy> or <move> operation.
* The <depend> task cache format has changed and all dependency information is
now stored in a single file.
|
|
|
|
(Commit approved by hubertf)
|
|
properly translated for buildlink.
|
|
|
|
|
|
(plus minor clean-ups: a paste-o and whitespace).
|
|
Handle the post 1.5 'arm32' -> 'arm' cleanup
|
|
|
|
|
|
|
|
so pkgsrc's libtool is invoked (this looks like an increasingly common
problem with up-to-date GNOME packages). Also change references to
*-config into ${*_CONFIG} so that the wrappers are used instead of the
original scripts (why is this such a common problem?!).
|
|
|
|
libintl.so won't pass the test for a GNU gettext, so building bison will
build the supplied gettext anyway, but there won't be a dependency on
gettext-lib in any case, so the bookkeeping is correct. If there isn't a
libintl.so in the base system, then bison will use the libintl.so installed
by gettext-lib, so again the bookkeeping will be correct.
So the short of it is that this is strongly buildlinked again.
|
|
* Bug fixes.
* Speed improvements.
* Upwardly binary compatible with 3.x versions.
* New CPU support: IA-64, Pentium 4.
* Improved CPU support: 21264, Cray vector systems.
* Support for all MIPS ABIs: o32, n32, 64.
* New systems supported: Darwin, SCO, Windows DLLs.
* New divide-and-conquer square root algorithm.
* New algorithms chapter in the manual.
* New malloc reentrant temporary memory method.
* New C++ class interface by Gerardo Ballabio (beta).
* Revamped configure, featuring ABI selection.
* Speed improvements for mpz_powm and mpz_powm_ui (mainly affecting small
operands).
* mpz_perfect_power_p now properly recognizes 0, 1, and negative perfect
powers.
* mpz_hamdist now supports negative operands.
* mpz_jacobi now accepts non-positive denominators.
* mpz_powm now supports negative exponents.
* mpn_mul_1 operand overlap requirements relaxed.
* Float input and output uses locale specific decimal point where available.
* New gmp_printf, gmp_scanf and related functions.
* New division functions: mpz_cdiv_q_2exp, mpz_cdiv_r_2exp, mpz_divexact_ui.
* New divisibility tests: mpz_divisible_p, mpz_divisible_ui_p,
mpz_divisible_2exp_p, mpz_congruent_p, mpz_congruent_ui_p,
mpz_congruent_2exp_p.
* New Fibonacci function: mpz_fib2_ui.
* New Lucas number functions: mpz_lucnum_ui, mpz_lucnum2_ui.
* Other new integer functions: mpz_cmp_d, mpz_cmpabs_d, mpz_get_d_2exp,
mpz_init2, mpz_kronecker, mpz_lcm_ui, mpz_realloc2.
* New rational I/O: mpq_get_str, mpq_inp_str, mpq_out_str, mpq_set_str.
* Other new rational functions: mpq_abs, mpq_cmp_si, mpq_div_2exp,
mpq_mul_2exp, mpq_set_f.
* New float tests: mpf_integer_p, mpf_fits_sint_p, mpf_fits_slong_p,
mpf_fits_sshort_p, mpf_fits_uint_p, mpf_fits_ulong_p, mpf_fits_ushort_p.
* Other new float functions: mpf_cmp_d, mpf_get_default_prec, mpf_get_si,
mpf_get_ui, mpf_get_d_2exp.
* New random functions: gmp_randinit_default, gmp_randinit_lc_2exp_size.
* New demo expression string parser (see demos/expr).
* New preliminary perl interface (see demos/perl).
* Tuned algorithm thresholds for many more CPUs.
|
|
* Use LIBTOOL_OVERRIDE instead of LTCONFIG_OVERRIDE so pkgsrc libtool is
actually invoked.
* Make sure that *-config scripts are invoked as $*_CONFIG so that the
wrappers can be used instead.
|
|
_never_ built, regardless of whether guile is detected or not.
* Strongly buildlinkify.
* Use LIBTOOL_OVERRIDE instead of LTCONFIG_OVERRIDE so that the correct
libtool is used during the build and installation.
* Add a pre-build target to symlink the libxml headers into ${WRKSRC} so
that references to <libxml/foo.h> will resolve to <gnome-xml/foo.h>. I
imagine that the sources will be fixed to refer to more recent header
locations in a future release of GConf.
|
|
purpose of x11.buildlink.mk is to force a package to be strongly
buildlinked, and that needs to happen on a package-by-package basis.
|
|
default by bsd.buildlink.mk.
* Use LIBTOOL_OVERRIDE instead of LTCONFIG_OVERRIDE to use the correct
libtool.
* Strongly buildlinkify.
|
|
|
|
|
|
http://ccache.samba.org/
|
|
on non-i386 because lcc appears to emit i386 assembly code even on my alpha.
There appears to be some amount of work involved with porting this to
non-i386.
|
|
dealing with info files i.e. explicitly specify which install-info
command should be used.
*Explicitly* do not use ../../mk/texinfo.mk to avoid running into chicken and
egg problem:
define INSTALL_INFO to the very install-info command this package provides
and use it in post-install *and* PLIST.
|
|
line length to BUFSIZ; it should now be possible to index the mozilla
source
|
|
|
|
|
|
|
|
|
|
Changes in version 1.0.1
========================
* Documentation improvements [Matthias Clasen]
* Use new API in FreeType 2.0.9 to greatly speed up coverage
calculation for Xft and FT2 backends.
* Some fixes for reading of OpenType tables.
* Add -D_REENTRANT to compilation when necessary. [Sven Neumann]
* Try and enable ClearType for Win32 backend. [Tor Lillqvist]
* Fixes for Korean glyphs in the basic-x shaper [Changwoo Ryu]
* Improvements for computation of average char/digit width [Sven Neumann]
* Bug and error reporting fixes
[Arnaud Charlet, Erewan Chenede, Mikael Hallendal, Stefan Israelsson,
Alex Larsson, Soeren Sandmann, Dan Winship]
|
|
Changes in version 1.0.1
========================
* Bug fixes, optimizations [Padraig O'Briain, Bill Haneman]
* Fix Win32 DLL name [Tor Lillqvist]
|
|
Overview of Changes in GLib 2.0.1
=================================
* Portability fixes for Sun's Forte compiler [Erwann Chenede]
* Performance improvements for GObject parameter lookup,
g_filename_to/from_utf8() [Alex Larsson]
* Actually check interface prerequisites [Matthias Clasen,
Miroslaw Dobrzanski-Neumann]
* Fix problem with glib-mkenums taking huge amounts of stack. [Owen Taylor]
* Fix g_signal_handlers_disconnect_by_func() for C++ [Damien Sandras]
* Fixes for g_log() and threading.
[Sebastian Wilhelmi, Miroslaw Dobrzanski-Neumann, Tim Janik]
* Make g_print(), g_printerr(), g_warning(), etc, convert from
UTF-8 to the encoding of the locale [Sebastain Wilhelmi, Tim]
* Fixes for GIOChannel on windows. [Tor Lillquist]
* Fix gsize/gint mismatches in giochannel.c [Miroslaw]
* Fix file descriptor leak in g_file_get_contents() [Matthias]
* Workaround iconv() problems on older Solaris [Lauri Alanko]
* Fix warnings with gcc-3.1 about asm const [Cody Russel]
* Minor bug fixes.
Other contributors: Hans Breuer, LEE Sau Dan, Sven Neumann, Salmaso Raffaele,
Akira Tagoh, Morten Welinder
|
|
* C Skeleton
Some projects use Bison's C parser with C++ compilers, and define
YYSTYPE as a class. The recent adjustment of C parsers for data
alignment and 64 bit architectures made this impossible.
Because for the time being no real solution for C++ parser
generation exists, kludges were implemented in the parser to
maintain this use. In the future, when Bison has C++ parsers, this
kludge will be disabled.
This kludge also addresses some C++ problems when the stack was
extended.
|
|
changes are:
2001-03-02 Kota ABE <k-abe@media.osaka-cu.ac.jp>
* include/setjmp.h: change the name of struct _sigjmp_buf into
struct _ptl_sigjmp_buf. FreeBSD 4.2R's <machine/setjmp.h> defines
the same struct.
2000-12-07 Kota ABE <k-abe@media.osaka-cu.ac.jp>
* src/io_api.c (read_timedwait): fixed: infinite loop when READ()
returns with EIO. (<hsaitoh@nts.ricoh.co.jp>).
* src/io_api.c (readv_timedwait): ditto.
|
|
other friendly people on port-macppc.
While I'm here rework to check hw.machine_arch first and then hw.model, so
we cna handle overlapping model matches between hw.machine_arch. We've not
hit any yet, but this is the right thing to do :)
|
|
libraries the following could occur
- a.out platforms: c++rt0.o would be missing and therefore
global contstructors would not work.
- ELF platforms libstdc++ and libm would not be explicitly
linked in.
This affected notably libqt and explains the build failure of kdelibs2 on
arm32.
Bump PKGREVISION and require it.
|
|
Set runtime link path _before_ linking with the various libraries.
|
