John Daggett (:jtd) Reporter
	Description • 15 years ago

After font system changes were checked in for 493280, synthetic bolding is broken (i.e. no bolding occurs for fonts that lack a bold face).  Synthetic italics are fine.

Steps:

Open URL in 2010-01-08-04-mozilla-central version of FF

Result: bolding does not show for the last three rows where fonts without bold faces are specified (or implied in the case of Japanese).

	Jonathan Kew [:jfkthame] Assignee
	Comment 1 • 15 years ago

I think the root of the problem is that the new font-list implementation does not create "fake" font entries for bold/italic versions where there are no actual faces in the family. This means that we have to set up GDI to do the synthetic bolding later, when the font is instantiated, and it's not passing this on correctly.

I don't want to revert to the old practice of creating fake font entries (besides the fact that it seems rather kludgey, that's probably one of the places where the new code has reduced RAM footprint significantly). So I'll trace how style information is handled and try to fix it properly in the font instantiation.

I'm surprised we didn't have a reftest that would catch this; obviously, we need to add one.

	John Daggett (:jtd) Reporter
	Comment 2 • 15 years ago

Things like synthetic rendering are hard to test because it requires some sort of consistency across testing platforms, from Win 2K to misc Linux flavors.  It might make sense to have a required set of testing fonts installed or load those fonts from a folder specified in a pref.

	Dão Gottwald [:dao]
	Comment 3 • 15 years ago

Shouldn't normal rendering always be different bold rendering, regardless of the platform, even if the requested font family is unavailable?

	John Daggett (:jtd) Reporter
	Comment 4 • 15 years ago

(In reply to comment #3)
> Shouldn't normal rendering always be different bold rendering, regardless of
> the platform, even if the requested font family is unavailable?

Yes, but most of the time you're seeing a normal bold face *not* synthetic bold/italics.  So you need to explicitly pick families that have just a regular face which varies across platforms and versions (i.e. W2K, XP, Vista, Win7).  That said, it should be possible to construct reftests that at least catch a problem in synthetic rendering for common environments.  It won't catch all situations but at least it will catch a majority of them.

	Dão Gottwald [:dao]
	Comment 5 • 15 years ago

Right, you'll need a couple of tests, each with a no-bold-face candidate for some platform.

	neil@parkwaycc.co.uk
	Comment 6 • 15 years ago

This also affects certain system fonts such as Microsoft Sans Serif.

Is it permissible to synthesize a bold version of a downloaded font?

	Jonathan Kew [:jfkthame] Assignee
	Comment 7 • 15 years ago

This patch should fix the issue on Windows. Includes reftests that should have detected the regression for us.

For some reason, the second reftest failed on Mac OS X when I pushed this to the tryserver; currently investigating this.

	Jonathan Kew [:jfkthame] Assignee
	Comment 8 • 15 years ago

The reftests pass for me locally on OS X; I don't know why #2 failed on tryserver. John, can you check if it works for you? If there's a real issue there, we should file it as a separate bug.

	Jonathan Kew [:jfkthame] Assignee
	Comment 9 • 15 years ago

I've modified the second reftest to prefer Apple Chancery on OS X, and this time it passed as expected on tryserver.

	Stuart Parmenter
	Comment 10 • 15 years ago

Comment on attachment 420965 [details] [diff] [review]
patch v1.1 - revised second reftest to use Apple Chancery on OS X

Won't this patch prevent you from actually getting real 800 and 900 weights properly?

	Jonathan Kew [:jfkthame] Assignee
	Comment 11 • 15 years ago

(In reply to comment #10)
> (From update of attachment 420965 [details] [diff] [review])
> Won't this patch prevent you from actually getting real 800 and 900 weights
> properly?

Errr.... yes, it would. :( Updated version attached that should handle this - WFM to access all weights of M+ fonts. Just pushed this to tryserver to re-check reftests.

	John Daggett (:jtd) Reporter
	Comment 12 • 15 years ago

Compare variations of bold, italic for different fonts to be able to catch busted synthetic bold/italic rendering.

	John Daggett (:jtd) Reporter
	Comment 13 • 15 years ago

Comment on attachment 420982 [details] [diff] [review]
patch v1.2 - fixed to allow faces with weight 800/900 to work

Looks fine.  I think reftests/font-matching would be a better place for the reftests than reftests/text.

	Jonathan Kew [:jfkthame] Assignee
	Comment 14 • 15 years ago

Pushed fix:  http://hg.mozilla.org/mozilla-central/rev/c7ff4d4a71a2
Extra tests: http://hg.mozilla.org/mozilla-central/rev/50566e7d7251