W3C home > Mailing lists > Public > public-test-infra@w3.org > January to March 2017

Re: Storage clearing

From: Quinten Yearsley <qyearsley@chromium.org>
Date: Thu, 23 Mar 2017 12:00:19 -0700
Message-ID: <CAFp4ESn85+s7e4LHJA=nz=4XAuLC1_SugLbuNJ1SDWQC-kL6Lw@mail.gmail.com>
To: Dirk Pranke <dpranke@chromium.org>
Cc: Philip Jägenstedt <foolip@google.com>, James Graham <james@hoppipolla.co.uk>, public-test-infra <public-test-infra@w3.org>, Jeff Carpenter <jeffcarp@chromium.org>, Joshua Bell <jsbell@chromium.org>
+jsbell since this is relevant to previous conversations

Original message for context:

I'm not opposed to the proposal of explicitly marking tests as requiring
storage to be cleared; implicit in this proposal is the idea that generally
most tests should not depend on have storage cleared, and if they do
require that, then they should explicitly say so. Is that right?

I know previously there was some discussion about whether storage should
always be cleared before the start of every test, but maybe that's not
necessary and would be slow?

Anyway, I'm not exactly sure now how we would implement it for Blink, but
I'm confident that we could, and the proposal sounds OK to me :-)

On Tue, Mar 21, 2017 at 10:28 AM, Dirk Pranke <dpranke@chromium.org> wrote:

> +qyearsley, jeffcarp ...
> Yes, we could implement something one way or another in Blink pretty
> easily, I'd think, but Quentin or Jeff could say for sure.
> -- Dirk
> On Tue, Mar 21, 2017 at 3:24 AM, Philip Jägenstedt <foolip@google.com>
> wrote:
>> On Tue, Mar 21, 2017 at 6:49 PM James Graham <james@hoppipolla.co.uk>
>> wrote:
>>> On 21/03/17 05:52, Philip Jägenstedt wrote:
>>> > Could that information be extracted out into the manifest? That seems
>>> more
>>> > straightforward than finding it and restarting.
>>> Yes, sorry the scheme I had in mind would have the manifest generator
>>> add this information to the manifest like we currently do for long
>>> timeouts, so that wptrunner or similar would have access to it before
>>> the test loaded.
>> Dirk, do you know enough about how we launch tests to say if this would
>> be workable in Blink? I'm guessing that we control the command line
>> invocation for each test, can avoid batching together tests that need a
>> clean slate, and run those tests using an empty profile?
Received on Friday, 24 March 2017 09:17:55 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 17:34:13 UTC