Would you mind looking into this drott?

Could this be related to  issue 132252 ? Do you have any additional info on whether the affected users are using font management software?

One report (with missing glyphs) indicates a case where duplicates of Helvetica were managed through font management software:
https://productforums.google.com/forum/#!topic/chrome/QZtI8VMjikU

Multiple users in https://productforums.google.com/forum/#!topic/chrome/8zEZvPty43c are reporting using font management software: Suitcase Fusion, Universal Type Client, and FontAgentPro. Some are saying that disabling and enabling duplicate versions of Helvetica helped.

Mostly it seems to affect Helvetica, and in some cases, Arial - probably related to multiple copies of a font with family name "Helvetica". 

U+0021 is exclamation mark, U+0022 is quotation mark, these are the two main glyphs we're seeing as incorrect replacements. In some cases '#', which is U+0023. So they are close in terms of codepoint value. 

The glyphs do seem to be shaped with the intended font, i.e. Helvetica or Arial, not with a fallback font, but they are spaced too far apart. A sequence of quote marks or exclamation marks would be spaced closely together.

One possible explanation is tha glyph id and glyph extents lookup gets mixed up and Skia returns incorrect glyph ID and a default extent. CC, fmalita@ & bungeman@.

If the glyph and extents lookup calls to HarfBuzzFace.cpp are coming in with a wrong codepoint, the expected artifacts would look different, the extents and advance information would be much narrower and the ", ! and # would not be spaced apart so far.

Text in Chinese websites do not show up in Chrome either.I also have Suitcase Fusion 6 installed.

diajiao@, do you see any duplicate fonts, or fonts overlapping between the system and your collection in Suitcase Fusion? Would you have some sample URLs where that happens? If you open the same page in Firefox, which font is used for the text you can't see in Chrome? Thanks!

I think what's happening has to do with this part of hb-coretext.cc, which assumes that postscript name identifies font faces uniquely, which obviously is not the case:

https://github.com/behdad/harfbuzz/blob/89137e325a19d62fb00979e1e8ab2ca59a2c1147/src/hb-coretext.cc#L887

As far as I know, there's no better way given the current CoreText API.

@drott@chromium.org I attached 2 examples, in both Firefox and Chrome:

http://baidu.com
http://us.weibo.com/gb

Both pages rendered okay in Firefox without deactivating my fonts. 

The text did show up in Chrome however, after I deactivated Times PostScript in my Suitcase Fusion. I also have Times TrueType as the system default, so problem solved after I deactivated Times PostScript. 

Deleted: Firefox_weibo.png 965 KB

	Firefox_weibo.png 965 KB View Download

Deleted: Firefox_baidu.png 181 KB

	Firefox_baidu.png 181 KB View Download

Deleted: Chrome_baidu.png 197 KB

	Chrome_baidu.png 197 KB View Download

Deleted: Chrome_weibo.png 995 KB

	Chrome_weibo.png 995 KB View Download

Thank you very much for the detailed reply. In order for us to reproduce it, could you tell me where you bought the Times Postscript version you have in your Suitcase Fusion?  Would you have a link?

@drott@chromium.org I would like to add that I have seen users reporting the same issue without having any active font manager software. The user did have FontAgentPro installed on the device. 

To be on the safe side we removed the FontAgentPro reinstalled Chrome browser and the issue was still appearing. Furthermore the issue was reproducible in Canary as well. 

We played around with the Sans-Serif default choice in the Chrome browser font menu (which was defaulted to Helvetica) changing to different types as TimesNewRoman, Arial, Courier and others. With Arial half of the page's content was displayed correctly. 

Encoding was set to Windows Western, we tried changing to UTF-8 and Macintosh Western with no success. Restarting the machine seems to temporarily fix the issue.

The page we tested with was @https://social.technet.microsoft.com/Forums/en-US/home

Fonts used on the page are - font-family: 'Segoe UI','Lucida Grande',Verdana,Arial,Helvetica,sans-serif;

Different elements on the page are all set to font-family: inherit so I was unable to determine the specific font used on separate page elements. OS is OSX, Chrome browser was at latest version and a fresh install of Canary reproduced the issue to a smaller extent. 

I can provide more information if you point me at what is needed. 

we will launch m50 stable soon, can a fix be made for it asap, and merge in by this Fri?

Re #14, langelov@ thank you very much for the detailed explanations and working with the user. Do I understand correctly, that the user had a font mgmt software installed, then quit it or uninstalled it, and the problem persisted?

I am assuming that the fonts that the font mgmt software registered to the system persisted despite the uninstall. 

Re #15: Tina, unfortunately, despite several days of attempting to reproduce the issue, I do not have a reproduction on any system. I cannot deliver a fix by Friday. 

One workaround that we have found working for some users: Check whether there are any fonts in the font mgmt software's collection (aside from the preinstalled system fonts) that have the same name as a system font, most prominent expamples: Helvetica, Arial, Times. If found, disable these duplicates, or remove these font files from the system.

As a speculative fix, we might attempt disabling font auto activation, see issue 453364. However, this is a long shot and we do not know whehter this will solve the issue, and I do not recommend merging that to stable without gaining experience with that approach in Beta.

I believe this is a new manifestation of the older  issue 132252 , hence removing the regression label.

drott@ Yes the user had font management software installed and we uninstalled it. The user claimed that they never used the software on their machine but had it installed as other team members are working with it all the time. That specific user said that their work does not need the usage of the software and thus they never activated it. (Other users are also seeing the issue and I asked them to try the workaround for affected users with fontManagement software in use - no info yet on results).

I will investigate into what files might be left on the machine in case the software was used at some point and try to uninstall them as well.

We did check the default OSX FontManager and made sure that the OS's default font is "inline" with the browser's default with no effect on the issue.

I will also try the provided workaround in #137 at https://bugs.chromium.org/p/chromium/issues/detail?id=132252

as it seems it fixed the problem for #157 (no font management software installed) and report back if any of the two has worked.

+ jshin@ to see if i18n team has any repro or suggestions.

drott@ seems to have done everything he could do so far. 

i18n and Chrome text team are in sync on this; no repro, no fix so far.

Quick update from the user side. 

Confirmed that /Library, /Library/Application Support, /System/Library/Extensions  are clear of any entries from the previously uninstalled FontAgentPro. Issue persists without the font management software.

#137 steps from  issue 132252  has fixed the issue for the moment. (https://bugs.chromium.org/p/chromium/issues/detail?id=132252)

Something I read somewhere in another thread which I hadn't considered and I'm now bering in mind when trying to troubleshoot this issue.

Be aware that if you are 'signed in' to Chrome as a particular persona, and have multiple personas, your Settings > Web Content > Customize Fonts will be different for each 'Person'. It helps to have them all set consistently so you can negate this functionality when you are switching personas.

I was the originator of #137 but was removed somehow?

https://bugs.chromium.org/p/chromium/issues/detail?id=132252#c137

How is this still an ongoing issue? This is Google we are talking about. I still find articles from 2012 about this. Time to switch back to Firefox I guess. 

I and many colleagues were having the exclamation point issue. I tracked it down to these fonts in our library:

CL Helvetica Condensed Light
Helvetica Fraction
Helvetica FractionBold

None of us has had the problem since removing these fonts.

The issues is analysed, root caused and fixed and is planned to be released in Chrome version 53. See  issue 132252  for details.

Do you know when version 53 will be released? In these days I have many many display problems: I dont' have listed fonts, but problem persists. Thank you very much

Please see: https://www.chromium.org/developers/calendar