Summary: | Layout Test http/tests/canvas/canvas-slow-font-loading.html is failing | ||
---|---|---|---|
Product: | WebKit | Reporter: | Robert Kroeger <rjkroege> |
Component: | Tools / Tests | Assignee: | Nobody <webkit-unassigned> |
Status: | NEW --- | ||
Severity: | Normal | CC: | cdumez, d-r, noel.gordon |
Priority: | P2 | ||
Version: | 528+ (Nightly build) | ||
Hardware: | Unspecified | ||
OS: | Unspecified |
Description
Robert Kroeger
2012-11-27 12:18:13 PST
Yes, I introduced this test in http://trac.webkit.org/changeset/135888. I'll take a look. I checked the diff from: http://build.webkit.org/results/Chromium%20Mac%20Release%20%28Tests%29/r135890%20%2829113%29/http/tests/canvas/canvas-slow-font-loading-diffs.html The output is a green rectangle which means that the Ahem font was fully loaded when drawing. slow-ahem-loading.cgi was supposed to delay the font loading long enough so that the fallback font is used. There are several possible explanations: 1. The Ahem font was cached in a previous test and the font does not need loading any more when canvas-slow-font-loading.html is run (At least some ports clear the cache between tests, I don't know about chromium). 2. The delay in slow-ahem-loading.cgi is not big enough, in which case the test would likely be flaky rather than consistently failing. Is the test failing if run alone? If not, that would hint that my font caching assumption is correct. Fails for me locally on chrome mac DRT when run alone. Output image is a green rectangle. (In reply to comment #4) > Fails for me locally on chrome mac DRT when run alone. Output image is a green rectangle. Does increasing the sleep value in slow-ahem-loading.cgi help? I increased it to 3 seconds, no joy. Test is also failing on the mac wk1 bots. http://build.webkit.org/builders/Apple%20Lion%20Debug%20WK1%20%28Tests%29/builds/4978/steps/layout-test/logs/stdio |