murmur-1.1.2 is out. Ebuild will compile ONLY murmur (the ebuild is a dual tarball for the client and the server) init file changed due to server command changed from "murmur" to "murmurd" sed statement to pre-setup the location of the murmur.sqlite database einfo to inform users howto initialize SuperUser passwd Reproducible: Always
Created attachment 141852 [details] media-sound/murmur-1.1.2 ebuild
Created attachment 141854 [details] media-sound/murmur-1.x /etc/conf.d/ file
Created attachment 141855 [details] media-sound/murmur-1.x /etc/init.d rc file
Created attachment 143803 [details] media-sound/murmur-1.1.2 ebuild SRC_URI shouldn't point do murmur, changed to mumble
Created attachment 146052 [details] improved ebuild for murmur-1.1.3 Version bump to 1.1.3 Added optional support for mysql Make depencency for sqlite optional, but default Added the scripts to the homedir of murmur (perhaps there is some better place for them?) Minor cosmetics Whats about updating this in tree too?
Created attachment 147192 [details] murmur-1.1.3.ebuild +manpages +logrotate With this ebuild i added the following features: - manpages get installed - replaced murmur.ini through murmur.ini.system, which is the intended ini file for system wide configuration. - added dbus config file - prevent scripts only used by mumble from being installed - added optional logrotate support - updated messages being displayed after emerging This ebuild needs the following files: murmur.confd murmurd.rc murmur-1.1.3-ini.system.patch murmur-1.1.3-logrotate.patch Maybe murmur.pl and weblist.pl should be added as cgi files in a cgi-bin directory. Therefore the webapp.eclass may be of help. But as murmur can be run without a webserver as well this should somehow be optional.
Created attachment 147193 [details] changes paths to be apropriate
Created attachment 147194 [details] changes some paths to reflect the install location
Matthias Fritsche (In reply to comment #8) > Created an attachment (id=147194) [edit] > changes some paths to reflect the install location > Thanks for your ebuild, it works on my hardened box since 2 weeks now. I just want to mention qt4 use flags should be more carefully inspected: *sqlite is not needed and don't suffice, sqlite3 is mantadory *ssl is mandatory too (so these 4 were sufficient for me: dbus sqlite3 ssl zlib) Thanks again, I hope this ebuild goes in portage soon.
Created attachment 149300 [details] media-sound/murmur-1.1.3.ebuild +updated dependencies I updated the ebuild according to El Goretto's suggestions, which includes the following: - removed sys-apps/sed dependency because this is a system dependency already. - changed dependency on qt with sqlite OR sqlite3 to a dependency on qt with sqlite3, because it wont work with sqlite only. - added dependency on qt with ssl support, which is needed as well.
Created attachment 150355 [details] media-sound/murmur-1.1.3.ebuild +fixed typo, removed patches - fixed typo (sytsem.d) - moved scripts to doc - removed username/pathname patches, now sed - added missing mumble-server-->murmur replacement - added dodir /var/lib/murmur for fowners etc.
Created attachment 150407 [details] fix "Useful scripts location elog" in 150355 attachment Just fix in pkg_postinst the location of the useful scripts.
1.1.4 released any chance for an updated ebuild in portage?
why is the sqlite flag turned on by default? There is an error statement that will cause the build to fail if either mysql or sqlite flags are not enabled. So defaulting the sqlite use flag on seems unnecesary. would there be any issues changing the ebuild to no longer have "+sqlite" in iuse?
The sqlite useflag is default on because an ebuild should by default work. The user should imho not have to change or add useflags to get an ebuild working. Without the sqlite useflag on by default, the ebuild would by default die.
murmur 1.1.4 is out, for my works just rename murmur-1.1.3.ebuild build works, but for some reasons I cannot fully test the server.
Created attachment 155199 [details] murmur-1.1.4.ebuild
Comment on attachment 155199 [details] murmur-1.1.4.ebuild - fixed typo (scipts) - fixed disabling of precompiled headers
the init script doesn't work for murmur-1.1.4... even when adding -fg at the end(--exec /usr/bin/murmurd -- -ini $MURMUR_CONFDIR/murmur.ini -fg &> $MURMUR_LOGFILE &) how do i know it doesn't work...simply because: 1)i can start a new murmur server and bind on the murmur port 2)i see nothing in ps aux,ps A etc...
it was the database that had wtong permissions...mabe you shoud check for the permissions in the init script...
Created attachment 158963 [details] new init script (works with logrotate) new init script (works with logrotate)
Created attachment 158965 [details] new confd
Created attachment 158967 [details] new logrotate script
Created attachment 158971 [details] new init, works with logrotate (sorry, i forgot something)
Created attachment 158973 [details] new init script (works with logrotate) big sorry, now its correct
Created attachment 158979 [details] 1.1.3 (little adjustments due to new init;should work with 1.1.4 after renaming) little adjustments due to new init scripts, should work with 1.1.3 and 1.1.4 BTW: I have problems with 1.1.4: heavy sound lags from to time
Created attachment 159023 [details] fixes new introduced logfile/logrotate problem, updated info
Created attachment 159033 [details] new logrotate conf
No offense to mattenklicker, but I still prefer the original work of jon, Markus & Thomas, and the update of Matthias Fritsche on the ebuild and especially the scripts (init and conf). They are cleaner IMHO. For example, I would consider that a single "conf.d" config file containing "MURMUR_ARGS="-v -ini /etc/murmur/murmur.ini" is really dirty, compared to the original one with separated variables. Or mattenklicker, can you explain your work and why you think there is something to improve here?
Created attachment 162191 [details] Updated ebuild A new ebuild, merging parts of bug #206792, mattenklickers's ebuild and Mathias'. Contains the logrotate and sed cleanup by mattenklicker, as well as his init/logrotate scripts and compile cleanup from the mumble ebuild. I skipped the mysql/sqlite useflag checks on Qt though, since for the split ebuilds it would have been a bit more complex to fix correctly.
Created attachment 162193 [details] Updated confd
Created attachment 162197 [details] Updated confd Previously added the wrong file...
Created attachment 162199 [details] Updated init script Uses mattenklicker's murmur.logrotate
Thanks for the ebuild, but the qt dependency is very annoying for a system without X. Maybe you can update the ebuild to use the "qt-static" flag as the skype ebuild [1] does. I'm sure there's more than one user interested in. [1] http://sources.gentoo.org/viewcvs.py/gentoo-x86/net-im/skype/skype-2.0.0.63.ebuild?view=markup
Created attachment 164884 [details] murmur-1.1.5.ebuild used Dennis' ebuild but - added dev-cpp/Ice dependency - added pch use flag (for hardened systems) - moved scripts to doc
Created attachment 164887 [details] init script --background is not needed anymore
murmur-1.1.6 compiles with murmur-1.1.5.ebuild if USE="-dbus" is set, murmur doesn't pull dbus as dependency and compilation breaks with errors about missing dbus header files.
Big thanks to everyone who contributed to this bug. I've just added murmur-1.1.6 to the tree, please test and report any issues.
Is dev-cpp/Ice really a mandatory dependency? I have a working ebuild with a USE controled dependency on ice and use ice || conf_add="${conf_add} no-ice" in src_configure()
(In reply to comment #39) > Is dev-cpp/Ice really a mandatory dependency? > I have a working ebuild with a USE controled dependency on ice and > use ice || conf_add="${conf_add} no-ice" > in src_configure() > That's correct, ICE is not a required dependency. I just asked one of the devs and he confirmed it.
Another mistake: The new init.d script lists dbus as needed. But Murmur will run perfectly fine without a dbus daemon running, it is only used. Solution: need dbus -> use dbus Together with the too hard dependency on Ice, so this bug should be *reopened*. Can the maintainer or reporter please do that?
1.1.6 compiles without boost, 1.1.5 did not. So I think boost isn't needed also.
(In reply to comment #42) > 1.1.6 compiles without boost, 1.1.5 did not. > So I think boost isn't needed also. > I also checked with the devs about that: Murmur can be built without ice and ice is the only thing in murmur that requires boost. If ice is included, boost is needed aswell. If ice is excluded, boost isn't needed either. So, if we introduce an ice useflag, the devs asked us to have it default-on, as that would be the standard behaviour the devs would anticipate. I agree, the bug should be reopened. The ebuild is good and mostly correct, but there are still some details to fix.
Thanks for your suggestions. I've commited the fixes/improvements, please emerge --sync in a few minutes. If you still encounter any issues please open a new bug.