>From a draft from 17 April. I think I wanted to check it again before sending, but never got to it.
On Tue, 27 Apr 2010 00:05:57 +0200 Benjamin Drung <[email protected]> wrote: > Am Sonntag, den 18.04.2010, 19:15 +0200 schrieb Erik Massop: > > I think I'll reject the supplied patch for Debian Bug 510375 > > (xmms2-launcher verbosity), and create a new patch introducing a > > -q/--quiet switch, so that no xmms2d-switch it shadowed. Is this okay > > with the Debian-people? > > I prefer a more silent xmms2-launcher by default. As stated in the bug > the launcher should only output error messages by default. Do you agree? Actually the bug says: "If you don't want those as default behaviour, command-line switches to achieve them would of course also satisfy me." > To avoid shadowing the xmms2d-switch let's rename the verbose parameter > for xmms2-launcher. I don't know what would be an appropiate name for a verbosity switch that cannot be --verbose and -v. Do you have any ideas? > > I'll reject the "Fix binary-or-shlib-defines-rpath Lintian > > warning"-patch because from xmms2 bugreport 1545 ("sid plugin fails to > > configure if resid is not in library searth path") it seems that the > > code that is removed, is needed for non-Debian systems. > > For which system is this needed? Can it be detected automatically? The default installation of libsidplay2 has a builders directory in '${libdir}'/sidplay/builders. Debian changed this to '${libdir}'/ (search for builders in [1]). The check in the wscript does seem a bit off though, as $ pkg-config --cflags --libs libsidplay2 returns -DHAVE_UNIX ${libdir}/libsidplay2.la (where ${libdir} is substituted) and waf will not set the LIBPATH_sidplay, since there are no -L switches (search for LIBPATH_ in [2]). A new test is needed. I'm thinking of just trying without rpath and if that fails setting rpath. > > According to http://www.gentoo.org/proj/en/qa/asneeded.xml a relatively > > recent GNU LD is needed to use the --as-needed switch and it will not > > work on FreeBSD. I'm rejecting the patch because failure to build will > > occur on non-Debian systems. > > Fine. Another solution that works everywhere is to fix these warnings: > > dpkg-shlibdeps: warning: dependency on librt.so.1 could be avoided if > "debian/xmms2-core/usr/bin/xmms2d" were not uselessly linked against it (they > use none of its symbols). ... > dpkg-shlibdeps: warning: dependency on libpthread.so.0 could be avoided if > "debian/xmms2-client-vis/usr/bin/xmms2-libvisual" were not uselessly linked > against it (they use none of its symbols). For a few of these warnings I found that the libraries were not explicitly mentioned. I think this is related to these lines in [3]: # add ancestors uselib too - but only propagate those that have no staticlib for v in self.to_list(y.uselib): if not env['STATICLIB_' + v]: if not v in self.uselib: self.uselib.insert(0, v) Could you maybe test with these lines commented out? Or add_as_needed defined in [3] should be used Regards, Erik Massop / nesciens [0] http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=510375 [1] http://patch-tracker.debian.org/patch/misc/view/sidplay-libs/2.1.1-7/configure [2] /wafadmin/Tools/config_c.py [3] /wafadmin/Tools/ccroot.py -- _______________________________________________ Xmms2-devel mailing list [email protected] http://lists.xmms.se/cgi-bin/mailman/listinfo/xmms2-devel
