[Lldb-commits] [PATCH] Documentation for test timeout

[Zachary Turner]

Just to be clear, im totally on board with the global timeout. It's the
per-test override I'm not crazy about, since it doesn't matter much for the
bot, and it's non-discoverable for the user.
On Thu, Dec 18, 2014 at 8:34 AM Vince Harron <vharron at google.com> wrote:

> > And in addition to both of those, we really should be either fixing the
> tests or disabling them. Now that this timeout is in, i have a feeling that
> fixing the hanging tests will be de-prioritized.
>
> Currently there are tests that only hang on the debian buildbot.  These
> hangs obscure all other test failures.  This patch was the only way to
> figure out what those tests are.  Unfortunately, they are not hanging
> locally for us and are therefore not at the top of our priority list.  We
> could mark them as XFail, but we still need the timeout in in case this
> happens again.
>
> Remember, without the timeout feature, if a single test hangs, you need to
> surgically delete the hung test or you lose all test results.
>
>
> On Thu, Dec 18, 2014 at 7:50 AM, Zachary Turner <zturner at google.com>
> wrote:
>>
>> My concern, and the reason I brought it up, is because it seems like
>> unnecessary complexity which only benefits a handful of people. I don't
>> like adding complexity that isn't useful for a wide audience.
>>
>> If this is just for making your own local test suite terminate in a
>> reasonable amount of time, then you can do that with an external shell
>> script which you run when you notice it's hung. If the issue is not
>> knowing
>> it's hung, then we should change the test runner to print its output as
>> the
>> test suite runs instead of at the end (we should do this anyway actually
>> because that's very useful).
>>
>> And in addition to both of those, we really should be either fixing the
>> tests or disabling them. Now that this timeout is in, i have a feeling
>> that
>> fixing the hanging tests will be de-prioritized.
>>
>> Anyway, this patch is going in (or has gone in) for now, but I really
>> don't
>> want to see this timeout logic get any more complicated in its current
>> form.
>>
>>
>> http://reviews.llvm.org/D6669
>>
>> EMAIL PREFERENCES
>>   http://reviews.llvm.org/settings/panel/emailpreferences/
>>
>>
>>
>
> --
>
> Vince Harron | Technical Lead Manager | vharron at google.com | 858-442-0868
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/lldb-commits/attachments/20141218/d29727c5/attachment.html>