If a call to `vcpkg_replace_string` makes no changes i.e doesn't
effectively replace a string, A warning is logged.
This should also help identify ports that no longer need these calls to
fix things in `.pc` files etc.
* [sqlite3] Add more features
* version
* Add dependency zlib for feature zlib
* version
* Remove -DSQLITE_API=extern
* version
* Add pre-defines to pkgconfig
* Add new option CFLAGS to get all clfags
* Get pkgconfig cflags, add sqlite3 feature dependency
* version
* doc
* Update to 8.1.1, cleanup patches
* Patch and use pc file installation from upstream
* Simplify dependency fixup
* Remove obsolete VCPKG_BUILD_SHARED_LIBS
* Revise feature and dependency interface
* Update versions
* Use all libs from proj.pc for gdal
* Set CMP0012 for dependency control
* Update versions
* Update versions
* Fix missing user32.lib when using libcrypto
* Update versions
* Fix missing user32.lib when using libcrypto
* Update versions
* Enable CI for libspatialite:x64-windows-static-md
* Use pkgconfig for nmake
* Use pkgconfig for autotools; fix mingw
* Install pc file for windows
* Add feature interface, make GPLv2+ features optional
* spatialite-tools need libspatialite[rttopo]
* Update versions
* Modernize portfile
* Update indentation
* Update to 5.0.1
* Don't enable legacy PROJ API on Windows
* Fix copyright
* Update versions
Co-authored-by: Billy Robert O'Neal III <bion@microsoft.com>
* [fontconfig] update fontconfig
* [libiconv] update to make
* [gettext] update to make
* [vcpkg_configure_make] Changes considering pure configure ports without the requirement to run autoconfig
* [fontconfig] make it work on windows
* Remove install of unofficial cmake config
* add function get_cmake_vars
* fine tuning.
* apply to make based ports.
* add log suffix on not windows platforms
* fix c&p error
* add previous LINK env
* setup env on windows and extract cpp flags correctly.
* update glib and libxml2
* fix windows regressions
* Apply suggestions from code review
* add windres wrapper to invoke RC.
* remove wrong fi
* fix libintl.
* try fixing uwp
* other things to update
* exiv2 fix
* libtool does not like -RTC1
* pass the /RTC<x> flag directly to the compiler
* trust cmake instead of adding extra flags
* fix expat
* fix iconv wrapper install
* change fontconfig intl linkage.
* add appcontainer back in which is required for UWP. Why is this not in the cmake flags?
* fix pkg-config in vcpkg_configure_make
* fix json-c pkg-config installation
* remove remnoved config
* comment out debug messages + bit of tuning
* finally fontconfig
* comment debug messages
* expat apply extra patches.
* Switch back to -E instead of -EP in cpp
* commit changes from fontconfig PR
* [expat] fix expat details
* Apply suggestions from code review
Co-authored-by: nicole mazzuca <mazzucan@outlook.com>
* cleanup docs
* update osx pipeline so that fontconfig actually builds
* fix expat the lazy way by using a higher commit than release
* fix a barage of regressions due to the use of unofficial targets
* [expat] use a higher commit hash with a better stabilized cmake build
* forgot freexl regression
* more unofficial fixes
* fix downstream expat usage
* fix wxwidgets
* fix gcdm regression
* [vtk] fix the regression due to expat
* add uwp cl flags back in. Somehow those are not set by cmake
* add a few other configure options for full gettext build
* fix cmake regression
* fix a few regressions
* fix static gettext build
* fix gettext static
* fix libxml2 wrapper for cmake
* some more regressions fixes
* add conversion from somelib.lib to -lsomelib
* add a few option to libiconv.
* get logs from CI
* add missing ar-lib wrapper
* add missing ar-lib wrapper
* add fatal error to gettext
* remove uuid from the list of LIBS since it seems to not exist in CI?
* small but important regex correction
* fix regex and add debug message for libs
* remove error and only build libintl.
* add uuid dependency to fontconfig in qt5-base
* osx install gettext for autopoint
* fix io2d regression by saying the port is broken .....
(which it is; upstream needs to learn proper cmake)
* restore the old libxml2 wrapper with minor modifications
* fix xmlsec regression
* install wrapper correctly
* try actual fixing io2d
* improve iconv wrapper
* add latest changes from update_fontconfig PR
* Apply suggestions from code review
first set which don't need special attention
Co-authored-by: ras0219 <533828+ras0219@users.noreply.github.com>
* Apply suggestions from code review
one more simple change
Co-authored-by: ras0219 <533828+ras0219@users.noreply.github.com>
* [x264] set env AS
* fix bugs due to refactor
* use subpath everywhere
* apply changes from CR
* fix fontconfig build.
* only change libs in static builds
* remove unnecessary lines 41 & 44
* remove flag transformation
* reintroduce the flag / to - transformation for MSVC
* trying to figure out autopoint issue.
using wrong (windows) find.exe instead of msys
* add correct working_dir to subpath
* escape stupid env paths.
* fix typo
* add findutils
remove debug messages
* add error if libtool chokes
* add file to msys
* pass lt_cv_deplibs_check_method=pass_all on windows
(couldn't get file.exe to work so that libtool correctly ids the passed libs)
* add bzip2 to msys
* reenable libtool check. Lets see if CI agrees
* unbreak linux
try to figure out where uuid should be on windows.
* add -L flag and help libtool ?
* try to get ci to find uuid.
* try to use cygpath
* update controls
* cleanup merge mistakes
* correct merge issues
* determine cmake vars if not done before.
* move adding of -l earlier
* more merge cleanup
* fix uwp builds by not transforming libs
* fix patches in io2d
* fix xz download error
* apply code review changes manually
* fix the typos left behind in CR
Co-authored-by: nicole mazzuca <mazzucan@outlook.com>
Co-authored-by: Billy Robert O'Neal III <bion@microsoft.com>
Co-authored-by: ras0219 <533828+ras0219@users.noreply.github.com>
* [LibLZMA] add a config-generated by CMake
* bump control files of LibLZMA and dependent ports
* [tiff] use proper liblzma target
[OpenCV] add an explicit dependency on LibLZMA for static linking
* [liblzma] fix header install path
* [LibLZMA] avoid using targets in old symbols
* fixes for windows-static and trying to see if CI is happier with a share/lib folder written in lowercase
* [LibLZMA] use only modern CMake language, remove patch in favour of target public definition
* [lzma] put symbols in cmake cache
* [libxmlpp] remove CMAKE_WINDOWS_EXPORT_ALL_SYMBOLS regression
* [lzma] fix header install path
* [liblzma] install wrapper to force config mode
* [liblzma] remove function check inside cmake config since we know it will pass
* [liblzma] wrapper fix
* [tiff,geotiff] general cleanup and patch fixes
* [libgeotiff] remove debug tools
* [tesseract] modernize
* [tiff] fix also tiff_library symbol
* [pdal,libgeotiff] better library integration
* [tiff] restore using unix i/o on UWP, since it was working... for sure win32 one cannot work
* [tiff] enable lzma also on uwp, since it works and is requested by many dependencies
* [selene] enable build on arm/arm64-windows, which was surely broken before
* [lzma] uniform naming with cmake 3.14
* [podofo] fix regression, it requires openssl which was disabled in dependencies
* [many ports] remove unnecessary mods
* [boost-iostream] chmod
* [openssl] fix regression due to missing architecture
* [podofo] fix regression on uwp due to missing include
* [libpq] explicitly fail on UWP, it should avoid being marked as regression
* [shiva] fix regression on linux
* [GEOS] Generate debug library names with `d` suffix
Currently geos.lib name is used for both, debug and optimised
GEOS libraries. This leads to situation when:
```
find_library(GEOS_LIBRARY_DEBUG NAMES geos)
find_library(GEOS_LIBRARY_RELEASE NAMES geos)
```
finds the same library for both
```
GEOS_LIBRARY_DEBUG=D:/vcpkg/installed/x64-windows/debug/lib/geos.lib
GEOS_LIBRARY_RELEASE=D:/vcpkg/installed/x64-windows/debug/lib/geos.lib
```
This is minimal patch that works around the problem.
Next, complete fix should be submitted to GEOS upstream,
preferably using exported targets.
* [libspatialite] Add missing geos suffixes in static builds
It has literals with latin chars in a C source.
Unfortunately VC++ recognize source file without BOM as in OEM code not in UTF-8.
It cause a strange behavior of VC++ compiler, and fails with
"epsg_inlined_prussian.c(685): error C2001: newline in constant."
It happens on non-latin Windows such as Japanese(CP932).
It also work when changing Windows configuration
"system locale for non-unicode applications" to latin one such as English(US)(CP432),
but it affects all applications and users in Windows system.
Signed-off-by: Hiroshi Miura <miurahr@linux.com>
