feat: xz-compress duckdb sources in the tarball (#530)

* ci: Check tarball size

* feat: xz-compress duckdb sources in the tarball

* Remove objects

* Windows too

* Some compilers

* sed

* Restrict

* gfind

* LF, not CRLF

.Rbuildignore

@@ -43,3 +43,5 @@ _files$
 ^compile_commands\.json$
 ^\.cache$
 ^tests/testthat/helper-rethrow\.R$
+^cleanup$
+^cleanup\.win$

.gitattributes

@@ -1 +1,6 @@
 src/duckdb/** linguist-vendored
+src/duckdb/**/*.c text eol=lf
+src/duckdb/**/*.cc text eol=lf
+src/duckdb/**/*.cpp text eol=lf
+src/duckdb/**/*.h text eol=lf
+src/duckdb/**/*.hpp text eol=lf

.github/workflows/install/action.yml

@@ -56,6 +56,8 @@ runs:
         echo "_R_CHECK_UNDOC_USE_ALL_NAMES_=true" | tee -a $GITHUB_ENV
         echo "_R_CHECK_URLS_SHOW_301_STATUS_=true" | tee -a $GITHUB_ENV
         echo "_R_CXX_USE_NO_REMAP_=true" | tee -a $GITHUB_ENV
+        echo "_R_CHECK_CRAN_INCOMING_=true" | tee -a $GITHUB_ENV
+        echo "_R_CHECK_CRAN_INCOMING_SKIP_LARGE_VERSION_=true" | tee -a $GITHUB_ENV
       shell: bash
 
     - name: Set environment variables (non-Windows only)

.gitignore

@@ -103,6 +103,7 @@ $RECYCLE.BIN/
 *.tgz
 src/**/*.o
 src/*.so
+src/*.dll
 /scripts/__pycache__
 /duckdb
 /src/Makevars.duckdb

cleanup

@@ -0,0 +1,8 @@
+#!/bin/sh
+
+set -ex
+
+git clean -fdx src
+cd src
+tar cvJf duckdb.tar.xz duckdb
+rm -rf duckdb

cleanup.win

@@ -0,0 +1,8 @@
+#!/bin/sh
+
+set -ex
+
+git clean -fdx src
+cd src
+tar cvJf duckdb.tar.xz duckdb
+rm -rf duckdb

configure

@@ -27,3 +27,10 @@ if [ -f "${DUCKDB_R_PREBUILT_ARCHIVE}" ] && tar -xm -f ${DUCKDB_R_PREBUILT_ARCHI
 else
   cp include/to-tar.mk Makevars.duckdb
 fi
+
+# The duckdb sources are xz-compressed in the tarball to keep it under 5000000 bytes.
+# This happens in the cleanup script.
+
+if [ -f duckdb.tar.xz ]; then
+  tar xJf duckdb.tar.xz
+fi

configure.win

@@ -33,3 +33,10 @@ if [ -f "${DUCKDB_R_PREBUILT_ARCHIVE}" ] && tar -xm --force-local -f ${DUCKDB_R_
 else
   cp include/to-tar-win.mk Makevars.duckdb
 fi
+
+# The duckdb sources are xz-compressed in the tarball to keep it under 5000000 bytes.
+# This happens in the cleanup script.
+
+if [ -f duckdb.tar.xz ]; then
+  tar xJf duckdb.tar.xz
+fi