- Force rebuild all consumers to fix potential miscompilations with
1.41.0
- Enable SOURCES by default. The sources are indexed by RLS and
required for it to function properly, so they should be available
by default. This also makes sure we test the option properly.
- Remove implied --config=config.toml from x.py args
- Switch to the upstreamed backtrace crate patches like rust-nightly
- Enable WASM by default [0]
- Strip libraries (D23650) [1]
- Simplify plist generation (D23735) [2]
Changes: https://blog.rust-lang.org/2020/02/27/Rust-1.41.1.html
Submitted by: mikael [0,1,2]
With hat: rust
Differential Revision: https://reviews.freebsd.org/D23835
The error message (in poudriere) is as follows:
====> Running Q/A tests (stage-qa)
Error: '/usr/local/bin/python' is an invalid shebang you need USES=shebangfix for 'share/vim/vim82/tools/demoserver.py.bak'
r522484 adds a new REINPLACE_CMD check which is enabled when DEVELOPER=yes. It
does not respect REINPLACE_ARGS (-i ''), thus runtime/tools/demoserver.py.bak is
generated. Vim installs everything in runtime/tools [1] which fails in stage-qa
for invalid shebang.
This workaround uses USES=python:env. It will replace the shebang with default
python (/usr/local/bin/python3.7).
[1] from src/Makefile:
# install the runtime tools
$(INSTALL_DATA_R) $(TOOLSSOURCE)/* $(DEST_TOOLS)
PR: 243961
Reported by: salvadore
Submitted by: sunpoet (myself)
Approved by: adamw (maintainer)
When python is built with DEBUG, vim fails to build. In fixing this
case, note that vim will now fail to build if python or vim is build
with DEBUG and the other isn't.
This also highlighted that python support is being kludged. The
PYTHON2 and PYTHON3 options have been combined into a single PYTHON
option that builds with the default python. Consequently, it is no
longer possible to build vim with both py2 and py3 support. I can't
name any modern high-usage python-based plugin that requires 2.7,
so hopefully this only messes up things for a very small minority.
Thanks to koobs for the prompt to combine these options. I've been
wanting to do this for a while now.
PR: 243606
Submitted by: novel
editing features, plus several not typically found in one's average text
editor (plugins being the most notable of these).
WWW: https://github.com/linuxmint/xed
This is not a resurrection of the previous `editors/xed' that was removed
in r506581 as abandonware.
- Force rebuild all consumers to catch regressions early
- Switch to cross-compiled (from amd64) bootstraps for all
architectures generated with the incoming lang/rust-bootstrap
- Update cargo-c to 0.5.2 to unbreak librav1e build
- Make use of regular MAKE_ENV/TEST_ENV in lang/rust
- Turn on RUST_BACKTRACE in lang/rust and USES=cargo to hopefully
produce more useful failure logs when something panics during
builds
Changes: https://blog.rust-lang.org/2020/01/30/Rust-1.41.0.html
Tested by: mikael, tobik
With hat: rust
Differential Revision: https://reviews.freebsd.org/D23385
Release 0.84.0:
core:
* Fix crash when converting from Unicode to ASCII-7
* Splash::scaleImageYdXu: Protect against crash if srcWidth is too big
* JBIG2Stream: fix potential crash in malformed documents
* JBIG2Stream: fix leak in reset() if called several times
* Internal code improvements
utils:
* pdfimages: Add error message if first page is larger then number of pages.
* pdfinfo: Improved paper size recognition
* pdfsig: Fix exit code when dumping signatures
* pdftocairo: Error out when even/odd selects 0 pages
* pdftohtml: Fix memory leak
* pdftoppm: Add an option to scale before rotate
* pdftoppm: Add -hide-annotations option
* pdftoppm: Error out when even/odd selects 0 pages
* pdftops: Improve -optimizecolorspace
qt5:
* Code cleanups
glib:
* Fix compiler warnings
Release 0.83.0
core:
* Improve when a file is recognized as Linearized
* Improve const-ness of the code
* Make code a bit more readable/maintanable
* Fix uninitialized memory uses in broken files
utils:
* pdffonts: Make code a bit more readable/maintanable
* pdftohtml: Make code a bit more readable/maintanable
qt5:
* Remove a bunch of unused internal functions
* trUtf8 -> tr (less warnings)
build system:
* make-glib-api-docs: switch to python3
PR: 243433
Exp-run by: antoine
Clang currently miscompiles emacs, resulting in broken builds for dependencies:
===> Building for treepy.el-emacs26-0.1.1_3
(cd /wrkdirs/usr/ports/devel/treepy.el/work-full/treepy.el-0.1.1 ; /usr/local/bin/emacs-26.3 -batch -q -f batch-byte-compile treepy.el)
*** Signal 6
Use GCC to workaround it.
PR: 243484
Approved by: jrm (maintainer)
GCC architectures have /usr/bin/gperf binary resulting in failing to configure:
checking for gperf... /usr/bin/gperf
checking whether gperf is new enough... configure: error: "GNU gperf 3.0.3" is too old or unrecognized, must be at least gperf 3.1
Put back GPERF to CONFIGURE_ENV to fix the build.
PR: 243215
Approved by: lwhsu (maintainer)
This update includes an Emacs major version change from 27.0.50 to 28.0.50.
Ports that may depend on editors/emacs-devel must chase this update
with a PORTREVISION bump.
Submitted by: ashish (maintainer)
Differential Revision: https://reviews.freebsd.org/D23012
Most systems are able to build vim in parallel most of the time.
The dependency chain in the Makefiles are a spaghetti mess and
concurrency is broken with some frequency, so it just isn't
feasible for it to be on by default.
It does shave quite a bit of time off the build (and I have had
.undef MAKE_JOBS_UNSAFE in Makefile.local for years), so now
there's an option. It's off by default.
PR: 243109
Reported by: glewis
graphics/gstreamer1-plugins-openexr: switch from C++98 to C++11 to
match new OpenEXR headers and unbreak compilation.
Bump PORTREVISION of all ports depending directly on either, based
on assessing INDEX-12 (bump_portrevision.pl -l (shallow) used).
Release Notes:
https://github.com/AcademySoftwareFoundation/openexr/releases/tag/v2.4.0
The update was deliberately deferred until after 2020Q1, and should
not be MFH'd back to it.
Add vscode to the portstree. the popular editor developed by Microsoft and
based on Electron.
Many thanks to the maintainer Hiroki for his very appreciated work on it.
libgoffice 0.10 has been introduced long enough ago 6 years ago, stopping support
for libgoffice 0.8, it is now more than time, to remove libgoffice 0.8 and only
keep 0.10.
Remove the optionnal goffice option for projects not supporting 0.10.
New versions of KDE applications landing in December
The release of new versions for KDE applications is part of KDE’s continued
effort to bring you a complete and up-to-date catalog of fully-featured,
beautiful and useful programs for your system.
Available now are new versions of KDE’s file browser Dolphin; Kdenlive, one of
the most complete open source video editors; the document viewer Okular; KDE’s
image viewer, Gwenview; and all of your other favorite KDE apps and utilities.
All of these applications have been improved, making them faster and more
stable and they boast exciting new features. The new versions of KDE
applications let you be productive and creative, while at the same time making
use of KDE software easy and fun.
We hope you enjoy all the novel features and improvements worked into all of
KDE’s apps!
Announcement: https://kde.org/announcements/releases/2019-12-apps-update/
Clang builds using -std=gnu++98 can use the STL headers in
/usr/include/c++/v1, but recent versions of the gcc headers forbid this.
As a fallback, OpenOffice tries to use the TR1 headers supplied by boost,
but recent versions of boost no longer have those headers. In theory
the gcc TR1 headers should work, but I was not able to make those work.
Solve this problem by doing gcc builds using the "bundled" version of
boost which is much older rather that boost from ports.
Patch the "bundled" boost to fix the compiler feature detection logic
to get rid of a bunch of compiler warnings.
The headers in the ports version of vigra has some C++11 stuff that
gcc also complains about, so use the "bundled" version of vigra for
gcc builds as well.
The icu patch may only be needed for non-default values of LOCALBASE.
Comment out an unused method in basebmp/inc/basebmp/packedpixeliterator.hxx
that contains some broken code that causes gcc to error out.
Add a missing include to l10ntools/source/inireader.cxx to resolve
a compile error.
Add a missing include to scripting/source/protocolhandler/scripthandler.cxx
to resolve a compile error.
No PORTREVISION bump since amd64 and i386 builds should be unchanged.
Tested by: Curtis Hamilton clhamilto AT gmail.com
Clang builds using -std=gnu++98 can use the STL headers in
/usr/include/c++/v1, but recent versions of the gcc headers forbid this.
As a fallback, OpenOffice tries to use the TR1 headers supplied by boost,
but recent versions of boost no longer have those headers. In theory
the gcc TR1 headers should work, but I was not able to make those work.
Solve this problem by doing gcc builds using the "bundled" version of
boost which is much older rather that boost from ports.
The headers in the ports version of vigra has some C++11 stuff that
gcc also complains about, so use the "bundled" version of vigra for
gcc builds as well.
The icu patch may only be needed for non-default values of LOCALBASE.
No PORTREVISION bump since amd64 and i386 builds should be unchanged.
Tested by: Curtis Hamilton clhamilto AT gmail.com
