I have the same effect starting very recently--I think Friday 6/30--*without* nth-of-type. Simply making the top and bottom individual borders of a single div the *same* alpha value, makes both of them not render at all. Same with left and right. As soon as they are made different, the borders appear.

I can't seem to connect to the link that you attached. Can you please attach a minified test case or link to a jsfiddle showing the issue? Thanks.

Sorry it would take too much time to set up a test case. It seems we
blocked external IP's. Can I give you access somehow?

2017-07-04 1:03 GMT+02:00 sh… via monorail <
monorail+v2.3439824626@chromium.org>:

Thank you for providing more feedback. Adding requester "shend@chromium.org" to the cc list and removing "Needs-Feedback" label.

For more details visit https://www.chromium.org/issue-tracking/autotriage - Your friendly Sheriffbot

If you could provide login details we could go from there, you can email it to one of our addresses on this bug.

Here's a jsfiddle showing a simple demonstration of the problem.

https://jsfiddle.net/wabvkpkm/

*Maybe* it's not the same root cause as the nth-of-type problem, but it probably is so I think you should take a look at it.

thanks, I've provided login details to the emailaddress provided in Comment 5. I have recreated the bug in this staging environment. Check out 

#usp .flex_column:last-of-type

in hp-main.css

Thank you for providing more feedback. Adding requester "dstockwell@google.com" to the cc list and removing "Needs-Feedback" label.

For more details visit https://www.chromium.org/issue-tracking/autotriage - Your friendly Sheriffbot

Chadd it looks like it's the same problem, The only thing I can remember (and can't replicate anymore for some reason, is that the borders become visible once I turn them on and off in the inspector.

I tried out the jsfiddle from #6 on Chrome for mac version 60.0.3112.50 but I'm not sure what I'm looking at. I see two divs with left/right borders, then two divs with top/bottom borders. What is the issue you're seeing there?

dstockwell, could you please see if you can repro using the login details provided as per #7? 

Thanks for trying on #10 there. There is an even more clear fiddle available: https://jsfiddle.net/wabvkpkm/2/

...but I think the problem could be different because my Chrome is this version on Windows: Version 59.0.3071.115 (Official Build) (32-bit)

I'll rescan the bug list and see if it has been filed separately for Windows.

yup, tested this already in browserstack and the problem is unique for the mac build. also doesn't occur on an earlier version!

So the problem is the borders disappear when the alpha's are the same.

Thank you for providing more feedback. Adding requester "meade@chromium.org" to the cc list and removing "Needs-Feedback" label.

For more details visit https://www.chromium.org/issue-tracking/autotriage - Your friendly Sheriffbot

This is a weird one!

I've updated the fiddle with some more test cases: https://jsfiddle.net/wabvkpkm/3/

Testing on Windows 10:

Chrome 59.0.3071.115 (Official Build) (64-bit) (cohort: Stable): BAD
Chrome: 61.0.3160.0 (Official Build) canary (32-bit) (cohort: 32-Bit): GOOD

Looks to be fixed on the most recent canary - we should bisect to find out what happened.

The good and bad build range didn't fall under the new bisect script to provide per revision bisect results, so providing chromium bisect results below

Bisect Information:
--------------------
Good build: 60.0.3096.0
Bad Build : 60.0.3094.0

You are probably looking for a change made after 470313 (known good), but no later than 470324 (first known bad).

Change Log URL: 
-----------------
https://chromium.googlesource.com/chromium/src/+log/1073d1c2e24a31bc6c34bbc6a6c198ae5d668086..175ceeff6a099a7923928ba8fcd58c6b80563251

Unable to find the actual suspect from the above CL, Could someone help us in assigning this issue to the concerned dev person.

Thanks!

Probably me. I am at least best placed to work on it.

First step is to redo the bisect.

Hooray, so it was a real bug! I feel important now. Thanks for handling the report in such a professional manner. Good work you guys. Love from NL

The bisect indicates M60, so adding a release blocker.

Unable to reproduce the issue on Macbook air 10.12.5 using latest beta #60.0.3112.72. Hence, removing the stable blocker for M-60.

bram@ - Could you please check the issue in latest beta #60.0.3112.72 and please confirm.

Thanks...!!

Note: I am able to reproduce on the reported version #59.0.3071.115. But it seems to be resolved on the latest M-60.

We've rolled out M-60 to most everyone, I think. And I cannot reproduce in later versions. So closing.