Investigate false First Contentful Paints from text rendering |
|
Issue descriptionFCP is fired when we render text. However, it may also fire when we render only whitespace, which really shouldn't be considered FCP. This bug aims to track efforts to reduce the false FCPs due to whitespace-only text. Issue 785344 is an example, but in general it's unclear which of the many callers of GraphicsContext::DrawText -> Font::DrawText actually require working through the text to check if there's only whitespace and passing on that information to the TextRun or some other member available in DrawText(). And even when there does exist a problematic case, it's also unclear whether the tradeoff between extra work to look for non-whitespaces vs. fixing the edge case is worth it.
,
Mar 27 2018
This should be marked Fixed, shouldn't it?
,
Mar 27 2018
Er, nevermind, this is for edgier edge cases than we've fixed. I'm almost inclined to mark this WontFix, as what we've got is probably good enough. Nicolas, what do you think?
,
Mar 27 2018
I think my initial CL had some room for improvement. But I think those are less common edge cases hence why this is P3. Maybe keep open for a while longer to see if I get some time to do it
,
Jul 16
Should we still keep this open?
,
Jul 16
Yes - currently we compute ContainsOnlyWhitespaceOrNbsp in the wrong place, which means not all trivial divs avoid FCP. For instance, '<div>  </div>' will trigger FCP and I don't think we want that. I think these are only edge cases but we may want to fix them eventually. I just don't know yet what's the correct place to compute ContainsOnlyWhitespaceOrNbsp.
,
Nov 14
Ping. Have you found the right place for ContainsOnlyWhitespaceOrNbsp?
,
Nov 14
I haven't had a chance to look into this again yet. |
|
►
Sign in to add a comment |
|
Comment 1 by tdres...@chromium.org
, Dec 11 2017Status: Assigned (was: Untriaged)