summaryrefslogtreecommitdiff
path: root/lang
AgeCommit message (Collapse)AuthorFilesLines
2012-01-09Add missing .endif.wiz1-1/+2
2012-01-08lang/gnat-aux: Fix introduced gnat.dg test failure on i386-netbsdelf-*marino1-1/+6
Previously the Ada testsuite was given unlimited stack resources for the x86_64 arch on NetBSD. Since all platforms now need unlimited stack resources to build gnat-aux due to the addition of <platform>-stdint.h header, this platform specific restriction on the Ada testsuite was removed. Unfortunately that resulted in a new stack test failure on i386 NetBSD platforms (gnat.dg/task_stack_align.adb execution test), so the original restriction seen in gnat-aux-20110627 was restored. Now i386 NetBSD once again pass all gnat.dg tests. This is strictly a testsuite issue so no PKGREVISION bump is necessary.
2012-01-08lang/gnat-aux: Upgrade to gcc 4.6.2, add Fortran, Objective-Cmarino16-664/+17538
Obvious additions: 1) Upgrades sync from gcc 4.6.1-RELEASE to gcc 4.6.2-RELEASE 2) New capability of building Fortran 3) New capability of building Objective-C 4) Building of all 5 languages (Ada,C,C++,ObjC,Fortran) now default 5) Fortran testsuite added 6) ObjC testsuite added Behind the scenes: 1) Previously GNAT-Aux was built from a custom-built tarball. Now real real gcc source files are used instead, but heavily patched. 2) The standard patch mechanism is not used. Composite diff files are generated by dragonlace.net and they are applied as needed, and are located in the "files" directory 3) This might be the only gcc that doesn't use the monolithic source tar ball. Depending on the options selected, the makefile updates its distfile list and only downloads what it needs, including testsuite files and dejagnu support. 4) All platforms are now built with unlimited stacksize command due to new issues with <platform>-stdint.h functionality. 5) All platforms use unlimited stacksize for Ada testing. Before it was limited to NetBSD x86_64. This may have introduced a failure on NetBSD i386 though. There were no other impacts according to the Ada testsuite results. 6) The PLIST automatic generation was significantly simplified, resulting in some variable deletion. 7) libstdc++ can't break testing now (forced to evaluate to true) 8) Unnecessary depends and USE_TOOLS removed 9) The includes-fixed directory is now removed from all platforms, arches now rather than problematic ones. It seems that it can only make no difference or cause problems, so no reason to keep it around. A) Unnecessary do-config phase "touches" removed. B) Several fixes added to diff patches to improve testsuite results on c, c++, and fortran for all platforms.
2012-01-04Missing dep on openssl, should fix linux builddholland1-2/+3
2012-01-04Missing openssl dep, should fix linux builddholland1-1/+3
2012-01-04Wrong distinfo was accidently generated.taca1-2/+2
Noted by David Wetzel on netbsd-users.
2012-01-04Update MASTER_SITES, HOMEPAGE.gdt1-3/+4
(Old versions do not resolve.) Also, add pointer (in comment) to debianized version on github. (no actual changes to the package; update to 0.11 is due but probably hard)
2012-01-03Add security fix for http://www.ocert.org/advisories/ocert-2011-003.htmltaca5-4/+65
from r321038 from PHP's repository. Bump PKGREVISION.
2012-01-031. Not MAKE_JOBS_SAFE.dholland4-11/+20
2. Use MMFLAGS instead of MFLAGS as the compiler flags make variable. The latter interacts somewhat poorly with make's own usage of the same identifier. Do this by SUBST at post-extract time so nothing ever sees the original form, and adjust patches to match. Does not build (it cannot parse NetBSD's stdlib.h) but no longer explodes randomly.
2011-12-30Use JAVA_ARCH not EMUL_ARCH in the de-install script.sbd3-5/+7
Bump PKGREVISION.
2011-12-28Update ruby18-base package to 1.8.7.357.taca4-13/+12
It contains security fix for CVE-2011-4815 (DoS). Wed Dec 28 21:34:23 2011 URABE Shyouhei <shyouhei@ruby-lang.org> * string.c (rb_str_hash): randomize hash to avoid algorithmic complexity attacks. CVE-2011-4815 * st.c (strhash): ditto. * string.c (Init_String): initialization of hash_seed to be at the beginning of the process. * st.c (Init_st): ditto. Thu Dec 8 11:57:04 2011 Tanaka Akira <akr@fsij.org> * inits.c (rb_call_inits): call Init_RandomSeed at first. * random.c (seed_initialized): defined. (fill_random_seed): extracted from random_seed. (make_seed_value): extracted from random_seed. (rb_f_rand): initialize random seed at first. (initial_seed): defined. (Init_RandomSeed): defined. (Init_RandomSeed2): defined. (rb_reset_random_seed): defined. (Init_Random): call Init_RandomSeed2. Sat Dec 10 20:44:23 2011 Tanaka Akira <akr@fsij.org> * lib/securerandom.rb: call OpenSSL::Random.seed at the SecureRandom.random_bytes call. insert separators for array join. patch by Masahiro Tomita. [ruby-dev:44270] Mon Oct 17 04:20:22 2011 Nobuyoshi Nakada <nobu@ruby-lang.org> * mkconfig.rb: fix for continued lines. based on a patch from Marcus Rueckert <darix AT opensu.se> at [ruby-core:20420]. Mon Oct 17 04:19:39 2011 Yukihiro Matsumoto <matz@ruby-lang.org> * numeric.c (flo_cmp): Infinity is greater than any bignum number. [ruby-dev:38672] * bignum.c (rb_big_cmp): ditto. Mon Oct 17 03:56:12 2011 Yusuke Endoh <mame@tsg.ne.jp> * ext/openssl/ossl_x509store.c (ossl_x509store_initialize): initialize store->ex_data.sk. [ruby-core:28907] [ruby-core:23971] [ruby-core:18121]
2011-12-25It is enough to use '=' to assign constant.taca1-2/+2
2011-12-25Fix some confusion about handling for RUBY_VERSION_DEFAULT and RUBY_VER.taca1-7/+4
Really fix the pkglint problem with ruby19-base and ruby193-base, and pbulk. Thanks to joerg@ to teache me pbulk-index!
2011-12-24Restore backwards compatibility removed as part of 1.65, the tree is notjoerg1-1/+8
ready yet (devel/ZenTest and others) and the tree is frozen.
2011-12-24Fix typo.joerg1-2/+2
2011-12-24* Remove backward compatibility handling for RUBY_VER; pkglint dosen't like it.taca1-8/+3
* Add one more PRINT_PLIST_AWK for GEM_HOME.
2011-12-23Avoid picking builtin SQLite.asau1-1/+2
2011-12-22Use COMPILER_RPATH_FLAG instead of -Wl,-R.wiz6-13/+13
Fixes OS X build. From Kai-Uwe Eckhardt.
2011-12-20+USE_GNU_READLINE= yes # rl_getc, rl_basic_quote_characters, ↵wiz1-1/+2
rl_insert_close
2011-12-20HOMEPAGE was added (PR 45726)cheusov1-5/+6
Fixes for pkglint warnings.
2011-12-19Fixes build of oss plugin for the case that ossaudio library is not required.obache3-14/+19
2011-12-19Not MAKE_JOBS_SAFE.dholland1-1/+3
2011-12-18Like lang/sr (this comes from the same group/project) there is no supportdholland1-1/+2
for x86_64.
2011-12-18Fix build on NetBSD/i386 5.99.58.ryoon3-3/+27
* Some pkglint. * Add options.mk and sqlite option (default: off). More options may be introduced.
2011-12-18Assorted fixes:dholland15-27/+574
- explain why we need post-extract chmods - sort PLIST - add patch comments - clean up some pkglint - fix a symbol name conflict with logf (from math.h + a gcc builtin) - fix some other bugs/issues found by gcc - add standard headers - remove some bogus BSD/System V include probing - probably fix gcc 4.5 build (not fully tested) - bump PKGREVISION
2011-12-18Add two "used by" line for lang/ruby193/Makefile andtaca1-1/+3
lang/ruby193-base/Makefile.
2011-12-17lang/schema48: Fix bad pthreads detection for DragonFlymarino1-2/+11
The schema48 configure schema has a pthreads test that can't be overridden. The problem is that it starts with -mt, and it thinks the test passes when in reality gcc complains. This commit does a post-patch inline replacement on the configure script to override the test, and to add -pthread to both $CFLAGS and $LDFLAGS.
2011-12-17lang/sr: Mask x86_64 platformsmarino1-1/+2
http://www.cs.arizona.edu/sr/impl.html: "SR does not run on 64-bit X86/AMD64 Linux". Indeed, the arch.h file has no provision for the x86_64 architecture. NetBSD x86_64 gets past the trap because it patched the arch.h file to alway define the arch. Configuring on DragonFly64 illustrates the arch is unsupported.
2011-12-17lang/swi-prolog-lite: Fix lang/swi-prolog-packages for DragonFlymarino2-1/+15
__DragonFly__ macro needed by sha1 for proper endian header. The problem manifested on lang/swi-prolog-packages which is based on swi-prolog-lite.
2011-12-17Update a component of pear package, Archive_Tar to 1.3.8.taca2-6/+7
Release date: 2011-10-14 23:02 UTC Release state: stable Changelog: * Fix Bug #17853: Test failure: dirtraversal.phpt [mrook] * Fix Bug #18512: dead links are not saved in tar file [mrook] * Fix Bug #18702: Unpacks incorrectly on long file names using header prefix [mrook] * Implement Feature #10145: Patch to return a Pear Error Object on failure [mrook] * Implement Feature #17491: Option to preserve permissions [mrook] * Implement Feature #17813: Prevent PHP notice when extracting corrupted archive [mrook] Bump PKGREVISION.
2011-12-17Change default PKGNAME scheme for PECL packages.obache1-2/+11
Drop ${PHP_BASE_VARS} from PKGVERSION by default. It used to be required to support multiple php version. But after PHP version based ${PHP_PKG_PREFIX} was introduced, such trick is not required anymore. In addition to this, such version name schme invokes unwanted version bump when base php version is bumped, plus, such version scheme is hard to use for DEPENDS pattern. To avoid downgrading of package using such legacy version scheme, PECL_LEGACY_VERSION_SCHEME is introduced. If it is defined, current version scheme is still used for currently supported PHP version (5 and 53), but instead of ${PHP_BASE_VARS}, current fixed PHP base version in pkgsrc is used to avoid unwanted version bump from update of PHP base package. With newer PHP (54, or so on), new version scheme will be used if it is defined. This trick will not be required and should be removed after php5 and php53 will be gone away from pkgsrc.
2011-12-17lang/squeak-vm: delint, add license, fix oss plugin for DragonFlymarino2-12/+19
DragonFly doesn't have the ossaudio library, so it won't build the oss plugin. The PLIST was adjusted accordingly. Pkglint hated the Makefile so it was cleaned up and a license entry (2-clause-bsd) was added.
2011-12-17Add missing mk/termcap buildlink.sbd2-3/+6
Bump PKGREVISION
2011-12-16lang/pnet: Mask for all DragonFly platformsmarino1-1/+2
It doesn't build on i386. When gcconfig.h is modified to recognize x86_64 platform, it breaks in the Boehm garbage collector. This is alpha-grade software from GNU that hasn't had a release in over 4.5 years. Frankly, I don't know how this abandoned project deserves a spot in pkgsrc.
2011-12-16lang/p2c: Fix WRKDIR permission, add licensemarino1-1/+8
For a reason I don't understand, the WRKDIR "work" directory ends up with file permissions of 777 and unknown user/group ownership. To make PKG_DEVELOPER=yes happy, changing the dir permission is enough.
2011-12-16lang/g95: Add PLIST.DragonFly similarly to PLIST.Linux (crt stuff)marino1-0/+5
2011-12-16On SunOS, don't ever override _XOPEN_SOURCE if it is already set.hans10-5/+80
Fixes build on SunOS with gcc>=4.6.
2011-12-16Expand comments. (No code change.)gdt1-4/+13
2011-12-16Clarify when distutils.mk should be used.gdt1-2/+3
2011-12-16Update ruby-rdoc package to 3.12.taca1-2/+2
=== 3.12 / 2011-12-15 * Minor enhancements * Added DEVELOPERS document which contains an overview of how RDoc works and how to add new features to RDoc. * Improved title for HTML output to include <code>--title</code> in the title element. * <code>rdoc --pipe</code> now understands <code>--markup</code>. * RDoc now supports irc-scheme hyperlinks. Issue #83 by trans. * Bug fixes * Fix title on HTML output for pages. * Fixed parsing of non-indented HEREDOC. * Fixed parsing of <code>%w[]</code> and other % literals. Issue #84 by Erik Hollensbe * Fixed arrow replacement in HTML output munging the spaceship operator. Issue #85 by eclectic923. * Verbatim sections with ERB that match the ruby code whitelist are no longer syntax-highlighted. Issue #86 by eclectic923 * Line endings on windows are normalized immediately after reading with binmode. Issue #87 by Usa Nakamura * RDoc better understands directives for comments. Comment directives can now be found anywhere in multi-line comments. Issue #90 by Ryan Davis * Tidy links to methods show the label again. Issue #88 by Simon Chiang * RDoc::Parser::C can now find comments directly above +rb_define_class_under+. Issue #89 by Enrico * In rdoc, backspace and ansi formatters, labels and notes without bodies are now shown. * In rdoc, backspace and ansi formatters, whitespace between label or note and the colon is now stripped.
2011-12-14Avoid using .warning make(1) directive which isn't supported on NetBSD 5.taca1-2/+2
Instead, append warning message to WARNINGS macro.
2011-12-14Don't use SIOCGIFHWADDR on SunOS.hans2-1/+17
2011-12-13Revert PYTHON_VERSIONS_ACCEPTED.joerg1-2/+1
2011-12-13lang/coq: Don't forget usmarino1-1/+2
DragonFly needs natdynlink enabled on all ocaml packages.
2011-12-13Upstream will not be moving to Python 3 in the foreseeable future. Seedarcy1-1/+3
http://blog.vrplumber.com/index.php?/archives/2540-More-porting-to-python-3-notes....html. Add license. The web page is unclear so I went with the most stringent BSD one. If I find out otherwise I will update it.
2011-12-13* Add RUBY_RAILS_SUPPORTED. Each Ruby on Rails related pacakge tellstaca2-34/+78
version(s) of acceptable Ruby on Rails. It almost replace previous RUBY_RAILS. * Add some definitions for Ruby on Rails 3.1.
2011-12-13Catch up to ocaml changes to enable natdynlink on NetBSD.dholland1-2/+3
2011-12-11Needs pkg-config.dholland1-2/+2
2011-12-08Force ExtUtils::MakeMaker to ignore installed packlist when creating newhiramatsu3-7/+19
packlist. It prevents generating wrong PLIST during make replace. Bump PKGREVISION.
2011-12-07Make this build on Mac OS X with ABI=64.yyamano2-4/+10
generated by cgit v1.2.3 (git 2.39.1) at 2025-02-10 13:18:34 +0000