RE: [dsdp-dd-dev] GDB ^error after ^ok

["Marc Khouzam" <marc.khouzam@xxxxxxxxxxxx>]

> On Friday 12 September 2008 18:04:54 Marc Khouzam wrote:
> > Hi,
> >  
> > In our code, in CLIEventProcessor.java, there is a comment saying:
> > 
> >         // GDB can send an error result following sending 
> an OK result. 
> >         // In this case the error is routed as an event.  
> 
> I presume you mean "^error" after "^done", as there's no 
> "^ok" response.

Right, my mistake.

> This cannot happen in current GDB, and even if it could 
> possibly happen is some version of GDB, it most probably some 
> 10-years-old gdb.
> 
> > another comment from CDI's MIErrorEvent is:
> >  
> > /**
> >  * (gdb)
> >  * &"warning: Cannot insert breakpoint 2:\n"
> >  * &"Cannot access memory at address 0x8020a3\n"
> >  * 30^error,msg=3D"Cannot access memory at address 0x8020a3"=20  */
> 
> This is valid output, but there's no "^done". 

I didn't know if this was the example that could create this case, so I
added
it here for information, thinking there might have be a ^done above,
which was
not copied into the comment.
 
> > I'm trying to figure out exactly how this case can happen (^error 
> > after ^ok) so that I can go on to ask how the new GDB 
> > (non-stop/multi-process) will be handling this case.
> 
> GDB, as result of any command, outputs either ^done, or 
> ^error. In some cases, it outputs ^running or ^connected 
> instead of ^done, but it cannot ever print "successful" 
> response followed by "failed" response, for the same command.

I was really hoping that was the answer :-)
So, I can do a bit of cleanup in GDB/DSF and not deal with this case.

Thanks!

marc