On Pi 5. August 2011 16:58:28 Martin Vidner wrote: > On Fri, Jul 29, 2011 at 02:39:21PM +0200, Stanislav Višňovský wrote: > > Hi! > > > > Here is updated patch with implemented some of the suggestions so far. > > > > Changes: > > - move almost all debugging flags to Debugger class instance, thus not > > influence memory requirements for non-debugger case > > > > - implemented remote debugging: > > /usr/lib/YaST2/bin/y2base --debugger-remote <YCP-client> <frontend> > > > > and client via > > > > /usr/lib/YaST2/bin/ycp-debugger <hostname> > > > > - use Y2DEBUGGER environment variable as also means to start local/remote > > debugger (set it to value "1" or "2") > > - provide better context information during tracing > > - do not allow to debug non-YCP code > > - refuse starting local debugging if socket already exists > > ah, but then it should clean it up on normal exit? > > > - fixed tracing if stepping from a breakpoint > > - some refactoring > > Committed! > http://svn.opensuse.org/viewvc/yast?view=revision&revision=65256 > With a small improvement to make CWM backtraces readable at all: > http://svn.opensuse.org/viewvc/yast?view=revision&revision=65257 > > yast2-core-2.21.5 submitted for next week's M4 of 12.1: > https://build.opensuse.org/request/show/78133
Thank you! > > > Feedback still welcome! > > Hmm, is it OK for CallFrame to keep an unmanaged pointer to params? I don't see right now how the pointer could become invalid while still on the CallFrame stack. > TODO: Move the socket to /var/run Good point. > TODO: Can we enable it by default for the installation? All you need to do is pass Y2DEBUGGER=2 to the kernel commandline. At least, this should work. > TODO: For the SLES11 SP2 installation? See above, I believe this should work out of the box. Stano -- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
