[webkit-reviews] review denied: [Bug 189724] REGRESSION: ( r231040 ) Layout Test http/tests/security/xss-DENIED-xsl-external-entity.xml is a flaky failure : [Attachment 363535] Patch

[bugzilla-daemon at webkit.org]

Alexey Proskuryakov <ap at webkit.org> has denied Alexey Proskuryakov
<ap at webkit.org>'s request for review:
Bug 189724: REGRESSION: ( r231040 ) Layout Test
http/tests/security/xss-DENIED-xsl-external-entity.xml is a flaky failure
https://bugs.webkit.org/show_bug.cgi?id=189724

Attachment 363535: Patch

https://bugs.webkit.org/attachment.cgi?id=363535&action=review




--- Comment #14 from Alexey Proskuryakov <ap at webkit.org> ---
Comment on attachment 363535
  --> https://bugs.webkit.org/attachment.cgi?id=363535
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=363535&action=review

If this is WebKit1 only and we honestly don't care, then maybe the right thing
is to keep the test marked as flaky. But a systemic problem with logging seems
like it is worth fixing at the root.

> LayoutTests/http/tests/security/resources/xsl-using-external-entity.xsl.php:3
> +header("Cache-Control: no-cache, must-revalidate");
> +header("Pragma: no-store");

I don't think "Pragma: no-store" exists, or does it? My comment about no-store
was with regards to Cache-Control. In any case, pragma should be unnecessary
for any browser engine made in this millennium I think.

> LayoutTests/http/tests/security/resources/xsl-using-external-entity.xsl.php:5
> +usleep(100000);

Ugh. In general, adding 100 ms timers doesn't fix flakiness on tests, because
there is so much CPU congestion that we can easily have multiple second delays.

> LayoutTests/http/tests/security/resources/xsl-using-external-entity.xsl.php:7
> +$contentType = $_GET['contentType'];

Where is this set?