On 3 Jun 2005, David Weeks <[EMAIL PROTECTED]> wrote: > Hello again. > > Continuing with my earlier question, I now ask how distcc and ccache > interact. I'm assuming at this point that the client/server > relationship of distcc solves potential recurssion problems within > distcc operations. In other words, I assume it is the distcc client > that distrubutes service requests to distcc daemons, yet the daemons > themselves NEVER distrubute service request, delivering to the compiler > instead, and routing the compilied binary back to the requesting host.
That's correct. > So I now wonder about the utilization of ccache. It seems to me that > the originating host ought to first que its compile list to its own > ccache, and where there is a miss, perhaps then que the job to a remote > host. When the remote host receives the service request, it too should > examine its ccache, and build only when necessary. Regardless of which > host eventually compiles an object, the originating host ALWAYS get's > the completed list of objects. That might be nice, but that's not what's done at the moment. The cache is strictly used before distribution. If you want a shared cache (which can be a good thing) you should use a network filesystem to sahre it. > Is there any state management of ccache > at this point? I don't know how ccache determines freshness, but I'm > sure it is some comparision of source code, say a check sum? Or is it a > make function? I don't know, but would like to. There is a hash; see the ccache documentation as Jeremy said. -- Martin __ distcc mailing list http://distcc.samba.org/ To unsubscribe or change options: https://lists.samba.org/mailman/listinfo/distcc