Dear Chromium Folk,

Today, something vaguely significant happened -- if you're into
numbers, statistic, and fixing layout tests. And you have to admit,
all of us are.

First, after a week of relentless triage, I finally drained the
Finders pool. "The Finders pool?" you ask, flashing back to your last
treasure hunt adventure. No, not that kind of finders pool -- well..
almost. See, over the time of project existence, we accumulated quite
a few layout test failures. As the Layout Test Task Force was
chartered, we split all of our failures into two pools:

1) the failures whose causes are known and are ready for fixing, or
the Fixers Pool
2) the failures that fail for reasons that are yet to be found, or the
Finders Pool

We even have this really neat dashboard
( -- thanks for building it, dpranke!)
that shows us these known knowns and known unknowns
Over the last few months, we've been diminishing the size of the Win
XP Release Finders pool, and last week I decided that the time has
come for it to say buh-bye. And id did! Now we _only_ have 159 bugs to
fix and we'll have 0 layout test failures. But at least we know what
to fix, right? I mean -- right? ;)

Oh -- and we have to keep this pool empty. Any creepage of unknown
failures will not be tolerated. Except when it's a flake -- we'll soon
have a separate triaging bucket for those, but for now all flaky tests
go into the Fixers pool.

My sincere apologies to tkent, jar, nsylvain, jungshik, sdoyon, ericu,
cevans, senorblanco and yuzo for inflicting the barrage of
rebaselining and expectation tweaking commits during their sheriffing
duty. That tree closing yesterday due to too many commits was
partially my fault.

Second, as a result of the aforementioned drainage, the Win XP Release
bot is finally showing under 300 failures! Woo hoo! This means that
over the last 4 months (or the entire length of its existence), the
LTTF eliminated over 500 test failures from the tree. That's big
stuff, folks.

Kudos to you all, LTTF-ers. I salute you on behalf of the entire
Chrome team and WebKit community, since a lot of tests fixed were not
specific to Chrome.

Anyhow, have a good night, y'all. Except for the WebKit gardeners --
who as it's widely known never sleep.

Chromium Developers mailing list: 
View archives, change email options, or unsubscribe:

Reply via email to