On 20.11.2020 10:05, Yasuhito FUTATSUKI wrote:
> On 2020/11/20 10:44, Yasuhito FUTATSUKI wrote:
>> On 2020/11/20 10:22, Branko Čibej wrote:
>>> On 20.11.2020 01:45, Yasuhito FUTATSUKI wrote:
>> Then if users want to rebuild bindings C source, they can do by using
>> --with-swig option in configure and doing
>> "make extraclean-swig-xx && make swig-xx".
> This is incorrect, because the build-outputs.mk generated by autogen.sh
> with --release option doesn't have rules to generate bindings C source
> files. So users need to execute autogen.sh without --release option in
> such case.
Right, and anyone who wants to generate the bindings sources with Swig
should be doing that in any case, so that's not a regression.
So it looks like your idea to detect "release mode" in configure is the
correct approach after all. But I don't like the idea of parsing
build-outputs.mk. How about something like this:
$ svn mv aclocal.m4 aclocal.m4.in
$ svn diff --no-diff-deleted --no-diff-added
Index: aclocal.m4 (deleted)
Index: aclocal.m4.in (added)
--- autogen.sh (revision 1883466)
+++ autogen.sh (working copy)
@@ -213,6 +213,14 @@ fi
# Produce ./configure
echo "Creating configure..."
+cp -f aclocal.m4.in aclocal.m4
+if test -n "$RELEASE_MODE"; then
+ cat <<EOF >>aclocal.m4
+# Generated by 'autogen.sh --release'
# If we have a config.cache file, toss it if the configure script has
--- . (revision 1883466)
+++ . (working copy)
Property changes on: .
## -14,6 +14,7 ##
Received on 2020-11-20 10:58:25 CET