On 2021/10/25 17:22, aisha wrote:
> another ping
>
> usockets - 0.8.1
> uwebsockets - 20.6.0
> purritobin - 0.6.7
> Additions to www/purritobin
> - auto cleaning of pastes
> - web portal for pasting
> - in-built http server, good for tiny pastes and testing setups
> - http header manipulation
>
> depends on a new port databases/lmdbxx, which is attached in the
> patch as well.
Please send new ports as a tar. Diff is harder to work with, and diff
mixing a new port (which many of us put in ports/mystuff/category/name)
with existing ports even harder.
> Cheers,
> Aisha
>
>
> diff --git a/databases/lmdbxx/Makefile b/databases/lmdbxx/Makefile
> new file mode 100644
> index 00000000000..b676b039ba3
> --- /dev/null
> +++ b/databases/lmdbxx/Makefile
> @@ -0,0 +1,41 @@
> +# $OpenBSD: $
> +
> +COMMENT = C++17 wrapper for Lightning Memory-Mapped Database
> +
> +VERSION = 1.0.0
> +DISTNAME = lmdbxx-${VERSION}
> +
> +CATEGORIES = databases
> +
> +GH_ACCOUNT = hoytech
> +GH_PROJECT = lmdbxx
> +#GH_TAGNAME = 1.0.0
> +GH_COMMIT = f1f00acf5f5045b43d7189c2b23ae5ca103a6faa
> +
> +HOMEPAGE = https://github.com/hoytech/lmdbxx
> +
> +MAINTAINER = Aisha Tammy <open...@aisha.cc>
> +
> +# Unlicense
> +PERMIT_PACKAGE = Yes
> +
> +COMPILER = base-clang ports-gcc
> +
> +NO_BUILD = Yes
> +
> +RUN_DEPENDS = databases/lmdb
> +
> +do-install:
> + cd ${WRKSRC} && \
> + mkdir -p ${PREFIX}/include && \
> + ${INSTALL_DATA} lmdb++.h ${PREFIX}/include
That dir is already created by ports infrastructure. Only one command so
the cd isn't needed, I'd go with
${INSTALL_DATA} ${WRKSRC}/lmdb++.h ${PREFIX}/include
> --- Makefile.orig
> +++ Makefile
> -@@ -1,60 +1,40 @@
> +@@ -1,72 +1,46 @@
> -# WITH_OPENSSL=1 enables OpenSSL 1.1+ support or BoringSSL
> -# For now we need to link with C++ for OpenSSL support, but should be
> removed with time
> -ifeq ($(WITH_OPENSSL),1)
> @@ -30,69 +30,85 @@ Index: Makefile
> - override CFLAGS += -DLIBUS_USE_LIBUV
> - override LDFLAGS += -luv
> -endif
> -+PREFIX ?= "/usr/local"
> ++PREFIX ?= /usr/local
> +LIBDIR ?= "$(PREFIX)/lib"
> +INCLUDEDIR ?= "$(PREFIX)/include"
>
> +-# WITH_ASIO builds with boot asio event-loop
> +-ifeq ($(WITH_ASIO),1)
> +- override CFLAGS += -DLIBUS_USE_ASIO
> +- override LDFLAGS += -lstdc++ -lpthread
> +- override CXXFLAGS += -pthread -DLIBUS_USE_ASIO
> +-endif
> ++PKG_CONFIG ?= pkg-config
> +
> -# WITH_GCD=1 builds with libdispatch as event-loop
> -ifeq ($(WITH_GCD),1)
> - override CFLAGS += -DLIBUS_USE_GCD
> - override LDFLAGS += -framework CoreFoundation
> -endif
> -+# OpenBSD specific library version
> -+LIBTARGET ?= libusockets.so.$(LIBusockets_VERSION)
> ++LIBTARGET = libusockets.so.$(LIBusockets_VERSION)
>
> -# WITH_ASAN builds with sanitizers
> -ifeq ($(WITH_ASAN),1)
> - override CFLAGS += -fsanitize=address -g
> -- override LDFLAGS += -lasan
> +- override LDFLAGS += -fsanitize=address
> -endif
Why remove all these WITH_XXX blocks? Having anything unnecessary in the
patch makes the patch harder to read, harder to merge with updates, and
makes it harder to review diffs to the port
> -# By default we build the uSockets.a static library
> -default:
> - rm -f *.o
> - $(CC) $(CFLAGS) -flto -O3 -c src/*.c src/eventing/*.c src/crypto/*.c
> +-# Also link in Boost Asio support
> +-ifeq ($(WITH_ASIO),1)
> +- $(CXX) $(CXXFLAGS) -Isrc -std=c++14 -flto -O3 -c src/eventing/asio.cpp
> +-endif
> ++all:
> ++ $(CC) $(CFLAGS) -c src/*.c src/eventing/*.c src/crypto/*.c
> ++ $(CXX) $(CXXFLAGS) -c src/crypto/*.cpp
> ++ $(AR) rvs libusockets.a *.o
> ++ $(CXX) $(CXXFLAGS) -shared -o $(LIBTARGET) *.o -Wl,-soname,$(LIBTARGET)
> $(LDFLAGS)
> ++ sed -e "s:@PREFIX@:$(PREFIX):" -e "s:@VERSION@:$(LIBusockets_VERSION):"
> libusockets.pc.in > libusockets.pc
> +
> -# For now we do rely on C++17 for OpenSSL support but we will be porting
> this work to C11
> -ifeq ($(WITH_OPENSSL),1)
> - $(CXX) $(CXXFLAGS) -std=c++17 -flto -O3 -c src/crypto/*.cpp
> -endif
> - $(AR) rvs uSockets.a *.o
lots of shuffling here that doesn't seem necessary, basically you just
want to remove -flto -O3 from the compiler lines don't you?
> -+COMMON_FLAGS = -DLIBUS_USE_OPENSSL -DLIBUS_USE_LIBUV -Isrc
> `$(PKG_CONFIG) --cflags libuv`
> -+override CFLAGS += $(COMMON_FLAGS)
> -+override CXXFLAGS += $(COMMON_FLAGS)
> ++install:
> ++ install -d "$(DESTDIR)$(LIBDIR)/pkgconfig" "$(DESTDIR)$(INCLUDEDIR)"
> ++ install -m 644 src/libusockets.h "$(DESTDIR)$(INCLUDEDIR)/"
> ++ install -m 644 $(LIBTARGET) "$(DESTDIR)$(LIBDIR)"
> ++ install -m 644 libusockets.a "$(DESTDIR)$(LIBDIR)/"
> ++ install -m 644 libusockets.pc "$(DESTDIR)$(LIBDIR)/pkgconfig/"
.....
> -+install:
> -+ install -d "$(LIBDIR)/pkgconfig" "$(INCLUDEDIR)"
> -+ install -m 644 src/libusockets.h "$(INCLUDEDIR)"
> -+ install -m 755 $(LIBTARGET) "$(LIBDIR)"
> -+ install -m 755 libusockets.a "$(LIBDIR)"
> -+ install -m 644 libusockets.pc "$(LIBDIR)/pkgconfig"
> -
patch block to add DESTDIR to these, can you avoid patching by simply
setting variables in FAKE_FLAGS? probably just setting PREFIX would cover
all of them
> clean:
> rm -f *.o
> rm -f *.a
> -- rm -rf .certs
> + rm -f *.so
> -+ rm -f libusockets.pc
> -+
> -+.PHONY: all install clean
> + rm -rf .certs
> ++ rm -f usockets.pc
> ++ rm -f hammer_test
but the file created *is* libusockets.pc? anyway ports doesn't call this
"make clean" there's no point patching it.
> --- a/net/usockets/patches/patch-Makefile
> +++ b/net/usockets/patches/patch-Makefile
> @@ -1,4 +1,4 @@
> -$OpenBSD: patch-Makefile,v 1.3 2020/12/11 22:33:22 sthen Exp $
> +: patch-Makefile,v 1.3 2020/12/11 22:33:22 sthen Exp $
broken rcsid line
