On Thursday 13 November 2008 02:19:57 Pedro Alves wrote:

> On Wednesday 12 November 2008 20:39:02, Vladimir Prus wrote:
> > This patch makes CLI 'detach' and MI '-target-detach' accept the PID
> > of the process to detach from.
>
> I see several issues with this patch:
>
>  - The target is not the right layer to do this.  Before you reach here,
>    you've already done things to the current inferior.  E.g.,
> target.c:target_detach will call remove_breakpoints before reaching
> remote_detach_1.  This means that, if you have e.g., selected inferior 1,
> and do detach 3, you'll remove breakpoints from inferior 1, and detach
> process 3.  Breakpoints are an example.  Other example is that before you
> reach the process stratum, you can pass by the thread_stratum, which again
> would do anything to the wrong detachee, since it's usually the
> process_statum that does the final real detach.  That's bad.

How about the attached -- where mostly everything is done on MI side?

- Volodya