On Friday 20 June 2008 06:48:52 Nick Roberts wrote:
>  > This patch fixes an issue in MI code that was present since at least 1999.
>  > We output ^running before even trying to resume the target, not to mention
>  > making sure the target is resumed. So, if resuming fails, we'd get ^running,
>  > followed by ^error, and I don't really know if current frontends will like
>  > it at all.
>  > 
>  > Now that we have observer for resume, and that observer is called after
>  > target is resumed, we can emit ^running from that observer. The immediate
>  > bonus is that ^running is now emitted for every command that resumes the
>  > inferior, even for CLI commands. Another (unexpected) bonus, is that since
>              ^^^^^^^^^^^^^^^^^^^^^
>  > now ^running and *running is output in a single place, we can produce them
>  > in consistent order.
> 
> I would like to see this patch committed!  I've not tested it but I should have
> Emacs working with MI shortly and then I can regularly test this and other
> changes.

I've checked in the following, which differs from original by extra test strictness.
Further, I've converted mi-async.exp to use the helper functions. Nick,
as it stands now it does not seem that mi-async.exp tests async behaviour
at all -- it merely changes that we get ^running for CLI commands, and we
get that in both sync and async mode. Do you think it worthwhile to rename
the test or move its content somewhere else?

- Volodya