On Mon, Nov 19, 2018 at 02:19:12PM -0600, Satya Mishra wrote:
> As requested in the bug report, here’s a patch that doesn’t hard-code the path to kf5-config, but instead assumes it’s in the path.
> > The test for kwallet assumes KDE5 when Qt5 present, but that is not
> > necessarily the case. For example CentOS7 has KDE4 and Qt5. The
> > correct test should test for kf5-config in addition to Qt5.
There are 3 ways that --with-kwallet can be used.
* --with-kwallet, everything is detected from the environment,
kf5-config should be in $PATH. AC_PATH_PROG() is used to find it.
* --with-kwallet=/basedir, kf5-config is expected to be under
* --with-kwallet=/incdir:/libdir, kf5-config is not used
> --- subversion-1.10.3/build/ac-macros/kwallet.m4 2018-10-24 11:50:47.000000000 -0700
> +++ subversion-1.10.3/build/ac-macros/kwallet.m4 2018-10-24 11:55:37.000000000 -0700
> @@ -45,7 +45,8 @@
> if test -n "$PKG_CONFIG"; then
> if test "$HAVE_DBUS" = "yes"; then
> + AC_CHECK_PROG(KF5_CHECK,kf5-config,yes)
> AC_MSG_CHECKING([for Qt])
> - if $PKG_CONFIG --exists Qt5Core Qt5DBus Qt5Gui; then
> + if test x"$KF5_CHECK" = x"yes" && $PKG_CONFIG --exists Qt5Core Qt5DBus Qt5Gui ; then
> AC_MSG_RESULT([yes, Qt5])
> qt_pkg_config_names="Qt5Core Qt5DBus Qt5Gui"
This patch should help with the first case above (aside from duplicating
the search for kf5-config). However, it seems like it would affect the
second case if /basedir wasn't part of $PATH.
Maybe some of the later logic that handles $svn_lib_kwallet's different
forms should be hoisted up here.
GPG Key: 4096R/91BF BF4D 6956 BD5D F7B7 2D23 DFE6 91AE 331B A3DB
Received on 2019-01-29 04:27:52 CET