Hi, On Friday 20 August 2010 03:40:38 pm viq wrote: > On Thu, Aug 19, 2010 at 12:31:29PM +0200, Sebastian Reitenbach wrote: > > On Thursday 19 August 2010 10:26:48 am viq wrote: > > > On Thu, Aug 19, 2010 at 01:04:20AM +0200, Sebastian Reitenbach wrote: > > > > Hi, > > > > > > > > Sebastian Reitenbach wrote: > > > > >viq wrote: > > > > >>On Sat, Jul 31, 2010 at 07:51:54PM +0200, Sebastian Reitenbach wrote: > > > > >>>viq wrote: > > > > >>>>On Sat, Jul 31, 2010 at 04:37:24PM +0200, Sebastian Reitenbach wrote: > > > > >>>>>Hi, > > > > >>>>> > > > > >>>>>forget the other mail, now the right tarball attached. > > > > >>>>> > > > > >>>>>Sebastian > > > > >>>>> > > > > >>>>>Sebastian Reitenbach wrote: > > > > >>>>>>viq wrote: > > > > >>>>>>>They released version 0.3.1, > > > > >>>>>>>http://spectrum.im/projects/spectrum/wiki/Changelog > > > > >>>>>>>But it hardcodes installation of spectrum.cfg to /etc > > > > >>>>>>>so that would need > > > > >>>>>>>to be patched, and it removes the sqlite schema, in repository > > > > >>>>>>>logs it says > > > > >>>>>>>"Remove unused outdated sqlite.sql schema." So for now > > > > >>>>>>>trying the posted > > > > >>>>>>>version, and it builds fine. Didn't get to trying it > > > > >>>>>>>yet, that will come > > > > >>>>>>>soon ;) > > > > >>>>>> > > > > >>>>>>With a lot of feedback from viq, updated to 0.3.1. > > > > >>>>>>I added a user _spectrum and group _spectrum so that the > > > > >>>>>>spectrumctl can change userid to that user. The homedirectory > > > > >>>>>> of that user has to exists, otherwise spectrumctl throws > > > > >>>>>> errors. The directories /var/spectrum and /var/log/spectrum > > > > >>>>>> are created from the port. I decided to put everything below > > > > >>>>>> /var/spectrum besides the logs. Hope that's fine. The example > > > > >>>>>> configuration file was updated accordingly. > > > > >>>>>> > > > > >>>>>>Maybe additionally a README.OpenBSD needed? > > > > >>>>>> > > > > >>>>>>cheers, > > > > >>>>>>Sebastian > > > > >>>> > > > > >>>>Following patch needs to be applied to PLIST, otherwise you > > > > >>>>end up with > > > > >>>>locale files being mode 640 root:_spectrum > > > > >>>> > > > > >>>>--- /tmp/net/spectrum/pkg/PLIST Sat Jul 31 13:51:54 2010 > > > > >>>>+++ pkg/PLIST Sat Jul 31 17:28:08 2010 > > > > >>>>@@ -40,6 +40,8 @@ > > > > >>>>@group ${SPGROUP} > > > > >>>>@mode 640 > > > > >>>>@sample ${SYSCONFDIR}/spectrum/spectrum.cfg > > > > >>>>+...@group > > > > >>>>+...@mode > > > > >>>>share/locale/cs/LC_MESSAGES/spectrum.mo > > > > >>>>share/locale/de/LC_MESSAGES/spectrum.mo > > > > >>>>share/locale/pl/LC_MESSAGES/spectrum.mo > > > > >>>> > > > > >>>>Another thing is, in > > > > >>>>lib/python2.5/site-packages/spectrum/doc/ there is > > > > >>>>interactive.py which after running spectrum gets bytecompiled > > > > >>>> creating interactive.pyc which remains after package removal. > > > > >>>> Solution I saw employed so far is to compile it at package > > > > >>>> creation time, > > > > >>>>so it ends up > > > > >>>>in PLIST. Don't ask me how to do it though ;) > > > > >>>> > > > > >>>>The mysql schema should probably be installed, for those who > > > > >>>>want to use > > > > >>>>it, maybe the munin monitoring script? Maybe some -extras > > > > >>>> package? > > > > >>> > > > > >>>There is another *pyc file that was not yet created. I now call > > > > >>>spectrumctl in post-install stage to create those pyc files. No > > > > >>> idea whether this is the right way to do, but it works ;) > > > > >>>I also included the mysql schema, since people might want to use > > > > >>> the mysql backend instead of the sqlite backend. > > > > >>> > > > > >>>Sebastian > > > > >> > > > > >>How about this? Command to compile .py files stolen from > > > > >>devel/py-gobject by ajacoutot@ ;) > > > > > > > > > >That's great, thanks for finding this. > > > > > > > > > >>Also moving sample config file and mysql schema to share/examples > > > > >>instead of share/doc as I believe that's where they belong. > > > > > > > > > >Yes, after sending the last update, I also thought about that this > > > > >is the right place for it. > > > > > > > > > >I only removed a superfluous rm -f in the post-install stage, > > > > >update attached. > > > > > > > > It did not made it to the release, and those people just released a > > > > new version. > > > > Major change is with the twitter transport, which I haven't tested > > > > due to lack of an account there, and added french translation, and > > > > not to forget the version number 0.3.1 -> 1.4.0 ;). > > > > > > It seems that twitter support requires installation of > > > pidgin-microblog, according to > > > http://spectrum.im/projects/spectrum/wiki/UpgradeNotes Which makes me > > > wonder a bit, maybe other pidgin protocol plugins could be made to work > > > with it too? Say, pidgin-tlen or pidgin-icb. But that would require > > > modifying dependencies to not include pidgin as such, but maybe just > > > libpurple, which from my reading of Makefile pidgin proper depends on > > > anyway. > > > > Therefore a dependency only to libpurple, and not to pidgin, is included. > > If I remember correctly, with the older versions, it did refused to > > compile when spectrum was unable to find libpurple. I don't know which > > transports need libpurple, and which need poco. > > I tried with ICQ, MSN, Yahoo, which worked for me. > > I know spectrum depends only on libpurple, I meant the protocol plugins, > say net/pidgin-tlen. And from quick glance at sources spectrum does need > libpurple to build, all protocols have (indirectly) #include "purple.h" > and all of them say for example: > const std::string gatewayIdentity() { return "icq"; } > const std::string protocol() { return "prpl-icq"; } > which to me looks like they use purple for that. > > So, again, my suggestion is that if the pidgin-microblog was to be > ported it should depend on libpurple and not on pidgin, and maybe same > treatment should be applied to already ported plugins - if it would be > possible to make spectrum work with them. Ah, now I got what you meant. I'll take a look.
Sebastian > > > Sebastian > > > > > Other than that I have not really tested it, I should really get to > > > it... > > > > > > > Maybe now OK to check in? > > > > > > > > Sebastian
