gclient: Keyboard interrupt should not cause request to send stack

gclient: Keyboard interrupt should not cause request to send stack

Patch-contributed-by: Alexey Marinichev <amarinichev@chromium.org>;

Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=37434

[Nasser Grainawi, 2010-01-28 19:50:33]

Should this catch any Keyboard interrupt while running a gclient command?

[M-A Ruel, 2010-01-28 19:52:24]

On 2010/01/28 19:50:33, Nasser Grainawi wrote:
> Should this catch any Keyboard interrupt while running a gclient command?

No, that's a good idea, it will reduce the noise sent to me. :)

[M-A Ruel, 2010-01-28 19:52:35]

lgtm

[amarinichev, 2010-01-28 19:55:40]

This is good enough to catch Ctrl-C.  For anything else we'll notify
Marc-Antoine.


On Thu, Jan 28, 2010 at 11:50 AM, <nasser@codeaurora.org> wrote:

> Should this catch any Keyboard interrupt while running a gclient command?
>
>
> http://codereview.chromium.org/553139
>

[Alexey Marinichev, 2010-01-28 19:56:52]

lgtm

[Nasser Grainawi, 2010-01-28 20:01:11]

So, this only works nicely for googlers, what about all of us non-googlers that 
don't want to see an ugly stack trace each time we hit Ctrl-C?

Alexey Marinichev wrote:
> This is good enough to catch Ctrl-C.  For anything else we'll notify 
> Marc-Antoine.
> 
> 
> On Thu, Jan 28, 2010 at 11:50 AM, <nasser@codeaurora.org 
> <mailto:nasser@codeaurora.org>> wrote:
> 
>     Should this catch any Keyboard interrupt while running a gclient
>     command?
> 
> 
>     http://codereview.chromium.org/553139
> 
> 

[M-A Ruel, 2010-01-28 20:03:52]

On 2010/01/28 20:01:11, Nasser Grainawi wrote:
> So, this only works nicely for googlers, what about all of us non-googlers
that 
> don't want to see an ugly stack trace each time we hit Ctrl-C?

breakpad is only enabled for googlers because I didn't want people to accept an
EULA to use depot_tools and we respect your privacy.

breakpad really just being triggered *after* the fact, the stack trace is
happening on python by design and the code is running after the stack trace has
been printed. There could be a ctrl_c shallower but that's outside of the scope
of breakpad.

[Nasser Grainawi, 2010-01-28 20:18:28]

maruel@chromium.org wrote:
> On 2010/01/28 20:01:11, Nasser Grainawi wrote:
>> So, this only works nicely for googlers, what about all of us 
>> non-googlers
> that
>> don't want to see an ugly stack trace each time we hit Ctrl-C?
> 
> breakpad is only enabled for googlers because I didn't want people to 
> accept an
> EULA to use depot_tools and we respect your privacy.
> 
> breakpad really just being triggered *after* the fact, the stack trace is
> happening on python by design and the code is running after the stack 
> trace has
> been printed. There could be a ctrl_c shallower but that's outside of 
> the scope
> of breakpad.

Ah, didn't realize it was still after it had been printed.

Would it be reasonable to do something like this?
http://lizards.opensuse.org/2009/05/04/supressing-keyboardinterrupt-traceback...

> 
> http://codereview.chromium.org/553139

[msb, 2010-01-28 20:31:02]

On Thu, Jan 28, 2010 at 12:18 PM, Nasser Grainawi <nasser@codeaurora.org>wrote:

> maruel@chromium.org wrote:
>
>> On 2010/01/28 20:01:11, Nasser Grainawi wrote:
>>
>>> So, this only works nicely for googlers, what about all of us
>>> non-googlers
>>>
>> that
>>
>>> don't want to see an ugly stack trace each time we hit Ctrl-C?
>>>
>>
>> breakpad is only enabled for googlers because I didn't want people to
>> accept an
>> EULA to use depot_tools and we respect your privacy.
>>
>> breakpad really just being triggered *after* the fact, the stack trace is
>> happening on python by design and the code is running after the stack
>> trace has
>> been printed. There could be a ctrl_c shallower but that's outside of the
>> scope
>> of breakpad.
>>
>
> Ah, didn't realize it was still after it had been printed.
>
> Would it be reasonable to do something like this?
>
>
http://lizards.opensuse.org/2009/05/04/supressing-keyboardinterrupt-traceback...
>
>
>
I kinda like getting the stack trace. I use a lot of tools written in Python
and I know that when I hit ctrl-C I'm going
to get a stack trace. Its useful sometimes when things are hung. Sometimes I
just hit ctrl-C to cancel an operation
I no longer want to happen (expensive/slow operation).

We could make this configurable but that just adds extra complexity and
means we won't get a stack trace when
things are hung. But too many ctrl-C are for the non-hung case that the
prompt is bothersome.


>
>> http://codereview.chromium.org/553139
>>
>
>