This is the mail archive of the
gdb-patches@sourceware.org
mailing list for the GDB project.
Re: [RFC v5] fix regressions with target-async
- From: Pedro Alves <palves at redhat dot com>
- To: Tom Tromey <tromey at redhat dot com>
- Cc: gdb-patches at sourceware dot org
- Date: Wed, 12 Mar 2014 19:27:06 +0000
- Subject: Re: [RFC v5] fix regressions with target-async
- Authentication-results: sourceware.org; auth=none
- References: <1394642546-10555-1-git-send-email-tromey at redhat dot com> <532096DD dot 2020201 at redhat dot com> <87wqfzkzm4 dot fsf at fleche dot redhat dot com>
On 03/12/2014 07:04 PM, Tom Tromey wrote:
>
> That approach is fine with me too. Then you can remove the odd linkage
> between the "maint" and target-async settings.
I would like remove that also for another reason:
The idea of "maint set target-async off" is to disable
_target_ async support, to emulate, say sync Windows testing,
on Linux.
Currently, on a non-async-capable target, "set target-async on"
itself does not error, but subsequent (CLI or MI) execution commands
fail with "Asynchronous execution not supported on this target.".
After the async-by-default series, "set target-async" becomes
exclusively an MI command. After "set target-async on",
MI execution commands are background execution commands.
Therefore, the combination:
"maint set target-async off" + "set target-async on"
should also result in the same exact error as today
you get when you try "set target-async on" against a
target that doesn't support async.
The linkage in v5 means that "set target-async on"
in the testsuite re-enables "maint set target-async on",
which defeats the purpose of the maint setting.
(btw, I'm considering adding "set mi-async", and making
"set target-async" a deprecated alias to that, for clarity)
--
Pedro Alves