This is the mail archive of the
gdb-patches@sourceware.org
mailing list for the GDB project.
Re: [PATCH] handle dprintf error more gracefully
- From: Antoine Tremblay <antoine dot tremblay at ericsson dot com>
- To: Pedro Alves <palves at redhat dot com>, <gdb-patches at sourceware dot org>
- Date: Wed, 4 Mar 2015 10:29:08 -0500
- Subject: Re: [PATCH] handle dprintf error more gracefully
- Authentication-results: sourceware.org; auth=none
- References: <1423585666-10300-1-git-send-email-antoine dot tremblay at ericsson dot com> <54EB45A4 dot 9030408 at redhat dot com> <54ECE3C5 dot 7020309 at ericsson dot com>
> Pedro:
> Alternatively, a new *error asynchronous MI event could be added,
> and then the frontend would be responsible for refresh all it's state
> when it got that, just like it should when it gets a synchronous
> ^error command result. (I haven't thought this options fully
> through.)
...
> Antoine:
So I don't think finish_thread_state_cleanup is a good place to fix the
issue.
I think it would be better to be directly in some catch probably around
handle_signal_stop.. not sure where exactly yet...
However I think the way of doing a observer_notify_normal_stop with
reason error is much better then what I had done at first !! :) And so use that.
Does it sound like a plan ?
Also, for the global *error.. I'm not sure, I think it's better even if
we give no guarantee on the state, that we try to advertise it to the frontend
as much as possible even in case of error... *stopped,reason does a better job
at that then *error... even if *error would be more general....
After thinking about this more, in order to be consistent, I think the
new *error asynchronous MI event is a better approach since we need to
make this consistent...
And having a special error handling case for each type of event did not
feel right...
I will rethink the patch with that in mind...
Regards,
Antoine