This is the mail archive of the
gdb-patches@sourceware.org
mailing list for the GDB project.
Re: [PATCH] Extra error message from update_watchpoint
- From: "Andrew Burgess" <aburgess at broadcom dot com>
- To: "Eli Zaretskii" <eliz at gnu dot org>
- Cc: "gdb-patches at sourceware dot org" <gdb-patches at sourceware dot org>
- Date: Wed, 30 Oct 2013 17:13:19 +0000
- Subject: Re: [PATCH] Extra error message from update_watchpoint
- Authentication-results: sourceware.org; auth=none
- References: <5260FD66 dot 7090506 at broadcom dot com> <52616D4B dot 3020209 at redhat dot com> <526FE5E9 dot 3000909 at broadcom dot com> <83ob68dmfb dot fsf at gnu dot org> <526FF27B dot 8090209 at broadcom dot com> <5270F23E dot 7010602 at broadcom dot com> <83a9hqepgt dot fsf at gnu dot org>
On 30/10/2013 3:41 PM, Eli Zaretskii wrote:
>> Date: Wed, 30 Oct 2013 11:49:18 +0000
>> From: "Andrew Burgess" <aburgess@broadcom.com>
>> cc: gdb-patches@sourceware.org
>>
>> OK, here are a few alternatives, feel free to pick your favourites:
>>
>> (1) The original:
>> error (_("Software read/access watchpoints not supported."));
>>
>> (2) Pedro's original replacement:
>> error (_("Hardware watchpoint support disabled. "
>> "See set/show can-use-hw-watchpoints."));
>>
>> (3) The original + why we can't use H/W watchpoints (bit long):
>> error (_("Software read/access watchpoints not supported, "
>> "re-enable hardware watchpoints using "
>> "\"set can-use-hw-watchpoints 1\""));
>>
>> (4) Same, but with a newline to keep it under 80 chars, not sure if
>> multi-line errors are acceptable though.
>> error (_("Software read/access watchpoints not supported.\n"
>> "Enable hardware watchpoints using "
>> "\"set can-use-hw-watchpoints 1\""));
>>
>> (5) Mention that H/W watchpoints are disabled, but not how to re-enable
>> them, though given the user has done the disabling this might be enough
>> to prompt them. This is single line, and just under 80 chars.
>> error (_("Software read/access watchpoints not supported, "
>> "hardware watchpoints disabled."));
>>
>>
>> My favourite is (5) at the moment, but I'll take whatever makes everyone
>> else happy :)
>
> None of the above really explains to the user why GDB is going to
> refuse to abide by her command.
>
> A good message should say something like "Cannot do SOMETHING because
> SOME-REASON."
>
> But I don't want to be in the position of blocking a commit due to
> something that is just MO, so feel free to ignore me.
I'd rather make you happy if possible, never hurts to suck up right ;-)
The following follows the structure you gave, I'd be happy to go with
this except that it's longer than 80 character... is that an issue?
"Can't set read/access watchpoint, software read/access watchpoints not
supported, and hardware watchpoints disabled (see set/show
can-use-hw-watchpoints)."
Thanks,
Andrew