README.copyright: New file

1 files changed, 125 insertions, 0 deletions

diff --git a/README.copyright b/README.copyright
new file mode 100644
index 0000000..c3cfe92
--- /dev/null
+++ b/README.copyright
@@ -0,0 +1,125 @@
+Starting from Scratch
+=====================
+
+First of course, unpack a clean source tree:
+
+    $ sudo rm -Rf tmp
+    $ sudo prokit build $root -bcT nop .
+
+Next get a list of all C source files (706 files as of version 1.32.0) that may
+be built into the `busybox` binary:
+
+    $ cd tmp/src/
+    tmp/src$ find * -name '*.[ch]' -a \! -path 'examples/*' \
+    > -a \! -path 'scripts/*' -a \! -path 'shell/*_test/*' | sort 0<../../.files
+
+The most common license notice text (in 359 files as of version 1.32.0) is
+"Licensed under GPLv2 or later, see file LICENSE in this source tree."  The
+second most common text (in 210 files) is "Licensed under GPLv2, see file
+LICENSE in this source tree."  Some (5) files contain both notices:
+
+    tmp/src$ cat ../../.files | xargs grep -Fl \
+    > 'Licensed under GPLv2 or later, see file LICENSE in this source tree.' \
+    > | xargs grep -Fl \
+    > 'Licensed under GPLv2, see file LICENSE in this source tree.'
+    archival/bbunzip.c
+    coreutils/cat.c
+    libbb/hash_md5_sha.c
+    loginutils/cryptpw.c
+    util-linux/mount.c
+
+Extract copyright comments from these (354 and 205) files:
+
+    tmp/src$ cat ../../.files | xargs grep -Fl \
+    > 'Licensed under GPLv2 or later, see file LICENSE in this source tree.' \
+    > | grep -Ev
+    > '^(archival/bbunzip.c|coreutils/cat.c|libbb/hash_md5_sha.c'\
+    > '|loginutils/cryptpw.c|util-linux/mount.c)$'
+    > | xargs ../../extract-copyright-comments \
+    > 1>../../.copyright-comments.gplv2orlater
+    tmp/src$ cat ../../.files | xargs grep -Fl \
+    > 'Licensed under GPLv2, see file LICENSE in this source tree.' \
+    > | grep -Ev
+    > '^(archival/bbunzip.c|coreutils/cat.c|libbb/hash_md5_sha.c'\
+    > '|loginutils/cryptpw.c|util-linux/mount.c)$'
+    > | xargs ../../extract-copyright-comments \
+    > 1>../../.copyright-comments.gplv2
+
+This can take around 45 minutes total on a 3.3-GHz Opteron 4238 CPU.
+
+Edit the files `copyright`, `.copyright-comments.gplv2orlater`, and
+`.copyright-comments.gplv2`.  Find all the unusual copyright and license notices
+(e.g. portions of code copied from elsewhere under other licenses), list them
+separately in `copyright`, and remove them from the `.copyright-comments.*`
+files.  Also remove comments where "(c)" appears as C code rather than as a
+copyright symbol (found in at least `shell/shell_common.c`).
+
+With only files with simply the common license notices remaining, extract the
+file names:
+
+    $ printf '%s, ' $(grep -v '^[[:space:]]' .copyright-comments.gplv2orlater) \
+    > | fold -s -w 81 | sed 's/ $//; $s/,$//'
+    $ printf '%s, ' $(grep -v '^[[:space:]]' .copyright-comments.gplv2) \
+    > | fold -s -w 81 | sed 's/ $//; $s/,$//'
+
+Add these file names and the copyright and license notices to `copyright`.
+
+Next extract copyright comments from the files found earlier to have both
+GPLv2-or-later and GPLv2-only license notices:
+
+    tmp/src$ ../../extract-copyright-comments \
+    > archival/bbunzip.c coreutils/cat.c libbb/hash_md5_sha.c \
+    > loginutils/cryptpw.c util-linux/mount.c \
+    > 1>../../.copyright-comments.gplv2both
+
+Edit the files `copyright` and `.copyright-comments.gplv2both`, moving copyright
+and license notices from the latter to the former.
+
+Now extract copyright comments from all the other (147 in version 1.32.0) files:
+
+    tmp/src$ cat ../../.files | xargs grep -FL \
+    > 'Licensed under GPLv2 or later, see file LICENSE in this source tree.' \
+    > | xargs grep -FL \
+    > 'Licensed under GPLv2, see file LICENSE in this source tree.' \
+    > 1>../../.copyright-comments.other
+
+And again edit `copyright` and `.copyright-comments.other`, moving copyright and
+license notices from the latter to the former.
+
+Also manually check the shell scripts under `applets_sh/` for copyright and
+license notices.  As of version 1.32.0 there are two files with no such notices,
+and since these scripts are embedded as-is into the `busybox` binary, it's
+unlikely that there will ever be any verbose comments.
+
+Upgrading to a New Upstream Version
+===================================
+
+Before upgrading, unpack a clean source tree of the old version:
+
+    $ sudo rm -Rf tmp
+    $ sudo prokit build $root -bcT nop .
+
+Next get a list of all C source files (706 files as of version 1.32.0) that may
+be built into the `busybox` binary, and extract all the copyright comments:
+
+    $ cd tmp/src/
+    tmp/src$ find * -name '*.[ch]' -a \! -path 'examples/*' \
+    > -a \! -path 'scripts/*' -a \! -path 'shell/*_test/*' | \
+    > xargs ../../extract-copyright-comments 1>../../.copyright-comments.old
+
+Now upgrade to the new upstream version (by editing `changelog` and downloading
+the new upstream source archive).  Unpack a clean source tree of the new
+version, and then get the copyright commands from all relevant C source files in
+the new version:
+
+    $ sudo rm -Rf tmp
+    $ sudo prokit build $root -bcT nop .
+    $ cd tmp/src/
+    tmp/src$ find * -name '*.[ch]' -a \! -path 'examples/*' \
+    > -a \! -path 'scripts/*' -a \! -path 'shell/*_test/*' | \
+    > xargs ../../extract-copyright-comments 1>../../.copyright-comments.new
+
+Finally compare the differences between the old and new copyright comments, and
+add these changes to the `copyright` file:
+
+    $ diff -u .copyright-comments.old .copyright-comments.new | less