This is the mail archive of the
gdb-patches@sourceware.org
mailing list for the GDB project.
Re: [MI/RFC] Emit ^running via observer.
> > The tests appear to fail now.
>
> Which tests?
All three in mi-async.exp, at the moment. I'll look at them more closely to
see if they are due to local changes.
> With unmodified CVS state, all MI tests pass for me both in sync
> and async mode. But indeed, if I make mi-async.exp not force async mode, it
> starts to fail....
mi-async.exp is a test for async mode, so is not expected to pass in sync mode.
> > SYNC:
> >
> > r
> > &"r\n"
> > ~"Starting program: /home/nickrob/myprog \n"
> > =thread-created,id="1"
> > ^running
> > *running,thread-id="all"
> > (gdb)
> > ~"\n"
> > ~"Breakpoint 1, main (argc=1, argv=0xbf961424) at myprog.c:146\n"
> > ~"146\tmain (int argc, char **argv) {\n"
> > *stopped
> > (gdb)
>
> ... in this way.
>
> > i.e no reason, frame, file, etc fields. It's important for the console in
> > a frontend that the CLI command generates the same MI output as the
> > corresponding MI command.
>
> Of course. This is pre-existing problem, though, and was present in gdb 6.8
> -- except that we did not output neither ^running nor *stopped at all -- so
> apparently making mi-async.exp not actually enable async mode is a bit
> premature yet. The problem, it appears, is that while the CLI command
> executes, 'uiout' is the CLI interpreter's uiout, not MI uiout. I'll
> probably won't have time to work on this in next month, so patches welcome.
It's not a problem if async mode becomes the default, which is my
understanding. In sync mode, t may be possible to switch to the underlying MI
interpreter while in "proceed" but I think that using async mode is cleaner and
I have no plans to make sync mode work in this way.
--
Nick http://www.inet.net.nz/~nickrob