- See https://www.mozilla.org/en-US/firefox/33.0/releasenotes/
- Fixes MFSA 2014-74-->82
- build with systemwide icu4c, workarounds #1064665 (and as a side
effect shrinks the libs/packages by 1 or 2 megs)
- add patch-toolkit_library_libxul_mk from bug #1036832 to workaround a
linking failure with our ancient binutils
- remove patch-media_libvpx_Makefile_in from #982693 fixing libvpx build
on i386, a variation of it was merged upstream
tested by bcallah@, benoit@ and Fabian Raetz, thanks!
- see https://www.mozilla.org/en-US/firefox/32.0/releasenotes/
- remove patch-toolkit_components_downloads_nsDownloadManager_cpp and
patch-xpcom_io_SpecialSystemDirectory_cpp fiddling with modes when
creating user dirs, they never made sense to me, and this is the wrong
place to touch this.
- remove patch-webapprt_moz_build, patch-webapprt_gtk2_Makefile_in and
patch-configure_in from #1012447, merged upstream
- See https://www.mozilla.org/en-US/firefox/31.0/releasenotes/
- Fixes MFSA 2014-56->66
- rm/add aboutcerterror/aboutCertError.xhtml patch, corresponding code
moved - i'm still pondering why we keep that patch..
- remove patch-browser_branding_official_Makefile_in, merged (#992613)
- remove chunk 1 of patch-configure_in, merged (#991686)
- improve skia/i386/sse3 patches from #1028827, for something simpler and
closer to what will be commited upstream
- fix README to point users to gstreamer1-plugins-libav instead of
gstreamer-ffmpeg, reminded by jturner@ a while ago
ok sthen@ jasper@ naddy@
in 30.0, see https://bugzilla.mozilla.org/show_bug.cgi?id=1028827.
Remove some "if build machine has SSE2/SSSE3 then don't bother with cpuid
feature checks at runtime" pieces.
ok jasper@ (landry is away for a few days).
- See http://www.mozilla.org/en-US/firefox/30.0/releasenotes/
- Fixes MFSA 2014-48->54
- Switch to use gstreamer 1.0 by default
- Add a bunch of patches from #992613, #991613, #1012447 to build
properly with gtk3
- Note that this fixes runtime on ppc which was broken since 29 (see
#961488), which was quite a hairy one
- Still running fine on sparc64 !
- See http://www.mozilla.org/en-US/firefox/29.0/releasenotes/
- Fixes MFSA 2014-34->47
- add spellchecker.dictionary_path as a default pref in all-openbsd.js,
pointing to ${LOCALBASE}/share/mozilla-dicts/ so that it looks for
systemwide dictionaries.
- Remove now useless patch-extensions_spellcheck_hunspell_src_mozHunspell_cpp
- Add patch for #990196, tries to use INSTALL_SCRIPT during build which
badly fails as user.
Note that starting with Fx 29, the sync protocol v1.1 used by
www/mozilla-sync still works but is not advertised anymore, and will be
phased out by v1.5 in Fx 31. www/mozilla-sync port will be updated in the
meantime, but users will have to reconfigure all their instances...
- See http://www.mozilla.org/en-US/firefox/28.0/releasenotes/
- Fixes MFSA 2014-15->32
- add an all-openbsd.js file to modify default preferences instead of
patching browser/app/profile/firefox.js like pigs. Install it in
browser/defaults/preferences/ so that it is effective by default for
newer profiles. Currently, we only use it to disable safebrowsing
feature..
- remove libffi patches, were merged upstream in #928390 and #928381
- See http://www.mozilla.org/en-US/firefox/27.0.1/releasenotes/
- Fixes MFSA 2014-01->13
- remove GStreamerLoader.cpp patch, merged in 927898 (or now useless)
- remove patch-gfx_layers_BufferUnrotate_cpp, merged in 929439
- remove patch-ipc_chromium_src_base_debug_util_posix_cc, merged
in 927810 (was p_psflags change in base)
Note: firefox 28 is due in ~10 days, but is depending on a SQLite
update in base, currently being worked on.
- see http://www.mozilla.org/en-US/firefox/26.0/releasenotes/
- fixes MFSA 2013-104 -> 117
- remove now useless sydney_audio_sndio.c file, nothing uses it anymore
- remove patch-ipc_chromium_src_base_dir_reader_bsd_h-from (#909005),
the getdents() patch has been merged upstream.
- see http://www.mozilla.org/en-US/firefox/25.0/releasenotes/
- fixes MFSA 2013-93 -> 102
- remove USE_GROFF (from naddy@)
- remove LDEP on gst-plugins-base, gst is dlopened at runtime now, and
there's a RDEP on gst-plugins-good anyway.
- add patch from #927898 to fix gst runtime dlopening (reminded by dcoppa@)
- add comment about ipc patches from #927810 & #909005
- remove 3 useless patches
- see http://www.mozilla.org/en-US/firefox/24.0/releasenotes/
- fixes MFSA 2013-76 -> 92
- remove BROKEN-sparc64, it now works thanks the awesome work of
martin@netbsd.org in bugs #871101 & #871555 among others i forgot
- add sparc64 & alpha (insane!) to MODGCC4_ARCHS
- remove patch-gfx_thebes_gfxPlatform_cpp, was merged upstream in
#722975 (and we dont build with system cairo anyway now)
- remove useless patch-parser_html_Makefile_in, maybe it made sense in
the gcc 3 / ffx 3.6 days but those are over...
(note: a spellchecker issue since fx 23 is known, and might be fixed
later on)
- use getdents(2) now that we have it; previously OpenBSD was special-cased
to use getdirentries(2) instead.
- interim fix: declare getdents() to allow firefox/seamonkey to build on
-current until consensus has been reached on return types/adding to system
headers.
discussed with landry@, clues from guenther@.
- see http://www.mozilla.org/en-US/firefox/22.0/releasenotes/ (ok,
webrtc isnt enabled yet by default on *BSD but we're working on it :)
- fixes MFSA 2013-49->62
- remove all patches related to sydneyaudio which got removed upstream,
sound uses libcubeb which has a native sndio backend since ages (thx
to ratchov!). also remove related post-patch target while here.
- add temporary workaround (TERM=xterm) for bug #878089
- switch i386 to use clang again, using --disable-debug-symbols to make
the code smaller (who said 'gross hack' ?)
- update patch-gfx_thebes_gfxPlatform_cpp to match what got commited
upstream in r52b02042b27f
- remove patch-js_src_vm_SPSProfiler_cpp which unbroke powerpc, finally
got commited in 22 branch
- rework fx-i18n's MESSAGE to be more in line with reality (discussed
with Marcus Merighi)
- See http://www.mozilla.org/en-US/firefox/21.0/releasenotes/
- Fixes MFSA 2013-41->48
- switch back again to use gcc 4.6 on i386, ld hits the memory exhausted
wall again with objects built with clang. doh.
- stuff shuffled around in PLIST
*BIG SCARY WARNING FOR PRIVACY-CONCERNED USERS*
Firefox 21 starts implementing something called 'Firefox Health Report',
which sends _anonymized_ data to mozilla servers about Firefox
performance.
This is opt-out - users will be prompted at first start if they want to
keep it enabled. This is not turned as an opt-in knob in the port on
purpose, because i got no feedback on this, and that will allow upstream
to know they have some OpenBSD users. This is of course subject to change.
See https://blog.mozilla.org/futurereleases/2013/05/14/firefox-heal-thyself/
& about:healthreport for more info about this.
- See http://www.mozilla.org/en-US/firefox/20.0/releasenotes/
- Fixes MFSA 2013-30->40
- switch back to use clang on i386, now that llvm got fixed there
- backport libffi i386 fix (TODO: switch to systemwide libffi if possible)
- add nicer fix to avoid installing the sdk lib/headers
- see http://www.mozilla.org/en-US/firefox/19.0/releasenotes/
- fixes MFSA 2013-21->28
- provides a builtin pdf viewer (pdf.js)
- remove patch-toolkit_system_gnome_nsGIOService_cpp (#805202), firefox
now properly links with gio instead of dlopening it
- remove patch-toolkit_xre_nsXREDirProvider_cpp and
patch-xpcom_io_nsAppFileLocationProvider_cpp, merged upstream (#803955)
- see http://www.mozilla.org/en-US/firefox/18.0/releasenotes/
- fixes MFSA 2013-14 -> 20
- build with clang on i386/amd64 and with gcc 4.6 on powerpc. Required
since upstream dropped support for gcc < 4.4.
- don't build against systemwide jpeg anymore since it now needs its
internal libjpeg-turbo.
- add a stub method to sydney_audio_sndio.c in post-patch. Cant add it
to the file in cvs directly otherwise it breaks other mozillas
- remove patch-ipc_chromium_src_base_atomicops_h, not needed anymore
ok sthen@
