From: didier <dga...@ma...> - 2009-02-25 01:03:28
|
Hi, Le dimanche 15 février 2009 à 18:52 +0100, Frank Lahm a écrit : > Hi, > > 2009/2/15 tim lindner <lin...@gm...>: > > Bolke de Bruin <bd...@gm...> wrote: > > > >> It is sometime ago that I had look at this code. In the first patch I > >> submitted I forget/messed up switch.c in that the aforementioned > >> function was actually not executed. Besides that to enable the patch > >> one has to add a configure variable when compiling. > > > > I just looked in the ViewVC web portal to the repository on sourceforge: > > > > http://netatalk.cvs.sourceforge.net/viewvc/netatalk > > > > I was trying to find the switch entry for FPSyncFork (0x4f, 79). As far > > as I could tell: It's not there. I think a patch was missed. The only patch I've committed is: > FPSyncDir patch (only) This is the working and cleaned up patch for FPSyncDir (command 78). It adds support for AFP3.2 as FPSyncDir is part of AFP 3.2. The patch has been tested on FreeBSD 7 & Linux 2.6. The patch itself makes volumes exported by netatalk supported by TimeMachine. Let me know in case of any questions. > Do you have a working FPSyncFork implementation? But thanks to Linux implementation I'm afraid it'd be useless. > Indeed, it look like that: > http://netatalk.cvs.sourceforge.net/viewvc/netatalk/netatalk/etc/afpd/switch.c?r1=1.13&r2=1.14 > > @Bolke: > if you could clean that up and send me patch againt current HEAD I'l > check it in. > > Imo you should set both function pointers to NULL in switch.c and add > them to auth.c as 3.2 calls: > http://netatalk.cvs.sourceforge.net/viewvc/netatalk/netatalk/etc/afpd/auth.c?r1=1.52&r2=1.53&pathrev=MAIN yes moving it to auth.c would be nice and we need a afp_null function without logging for at least 4C (no reason to spam log). Didier |