791048 - table cells disappear in FF15 on Mac

Status: RESOLVED WORKSFORME

(Core :: Layout: Tables, defect)

Description

[Chris Plummer]

Attachment: An example HTML document that illustrates the problem.

I have a widget in my web application that emulates a table that has a "non-scrolling" header by making two tables: one for the body and one for the header.  Each table is styled with table-layout: fixed and is wrapped in a div.  Each of these "inner" tables is held in an outer table so that they appear on top of each other - I've attached a sample HTML document with CSS to show it.

I use a CSS hover style on the rows to mark them as "clickable" in the application, with checkboxes in the first column to give users multiple ways to select one or more rows of the table.

In FF15 on Mac, the table cells at the top disappear as I hover over them (I can always reproduce it if I hover from top to bottom, then bottom to top, and so forth).  If I remove the '*' rule in my CSS, everything works fine again.  In my application, I have numerous tables like this, and each one has to be fixed differently - sometimes, I have to set the line-height on the table rows, and sometimes I have to set the height explicitly on other elements "above" the table in the DOM structure.  In one case, setting the 'height:15px' on a <legend> for an encompassing <fieldset> fixes the issue.  It seems like there is some kind of calculation error happening and the renderer doesn't know how to set the background color of my table rows correctly.

Of course, I expect the table rows to not disappear - the background color should show up correctly and render my table rows when I hover over them.

This works correctly in FF10 and FF4 on Mac OS X, but not in FF15 or later (I tried the 16 beta and the nightly 18 alpha build as well, but they both failed).  I deleted and re-installed FF15 also, just to be sure.

Comment 1

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

I don't think I'm seeing the problem, but I'm also not sure that I'm doing the right thing.  Could you describe more precise steps to reproduce the problem using the attached testcase, what you expect to see, and what you see instead as a result?

Comment 2

[Chris Plummer]

Hey David,

Please tell me if any of this isn't clear enough and I'll elaborate more.  I've also attached a PNG screenshot.

1) Open Firefox 15 on Mac OS X (I've reproduced on 10.7.4 and 10.8)
2) Clear the browser cache (Tools->Clear Recent History, uncheck all except 'Cache' and choose 'Everything' from 'Time Range to Clear').
3) Paste the attachment URL for this bug into the browser - 'https://bug791048.bugzilla.mozilla.org/attachment.cgi?id=660915'.
4) Type 'Enter' to load the page.
5) Move the mouse over the first row in the table with a checkbox in the first column.  Its contents disappear.
6) Move the mouse over the second row in the table with a checkbox in the first column.  Its contents disappear, and the first row's contents stay invisible.
7) Move the mouse over the third row in the table with a checkbox in the first column. It is highlighted correctly with a light-blue color and its cells appear correctly.  The second rows contents appear again, but the first row's contents remain invisible.
8) Move the mouse over the fourth and fifth rows in the table with a checkbox in the first column.  They each behave as the third row did above, with a correct blue highlight on them.  The first row remains the only invisible row.

Comment 3

[Chris Plummer]

Attachment: A screenshot of step 7 showing the problem.

Comment 4

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

Did you try on other platforms and find that it only happens on Mac?  Or did you only try on Mac?

Comment 5

[Chris Plummer]

I have a Fedora 17 box with FF 15 and a Windows XP laptop with FF15 - both of those configurations worked correctly.

Comment 6

[Mats Palmgren (inactive)]

I can reproduce with Fx15.0.1 and Nightly on OSX.  Works fine on Linux64.

Comment 7

[Benoit Girard (:BenWa)]

We're actually also seeing this problem with the profiler front-end but it reproduces less reliably then this example.

Comment 8

[Benoit Girard (:BenWa)]

Last good nightly: 2012-04-10
First bad nightly: 2012-04-11

Pushlog:
http://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=7702bca6b64d&tochange=e636439e342f

Comment 9

[Benoit Girard (:BenWa)]

Regression: http://hg.mozilla.org/mozilla-central/rev/25a13d26509d

Comment 10

[Jonathan Kew [:jfkthame]]

I think bug 830406 is most likely this same issue, and describes another scenario where it can easily be reproduced (in the Preferences dialog).

Comment 11

[Benoit Girard (:BenWa)]

I don't think we should ignore this bug. Requesting tracking.

Comment 12

[Robert O'Callahan (:roc) (email my personal email if necessary)]

I just tried to reproduce this following the steps of comment #2 and failed.

Comment 13

[Benjamin Kerensa [:bkerensa]]

Benoit, it seems Robert is unable to reproduce this and you have said you have not been able to reproduce this as reliably as the reporter. Is this something we would really consider blocking release on?

Comment 14

[Benoit Girard (:BenWa)]

Alright cleopatra doesn't use tables anymore. I really worry that some times will consistently hit this bugs and <table> is still an important web features but I don't have time to chase down what happened here.

Ideally we would have someone that would verify the above test case and perform a regression window to find when it broke and see if/when it got fixed.

Comment 15

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

Could this be specific to HiDPI displays?  (Are the people who see it on Mac all using such displays?)

Comment 16

[Chris Plummer]

Hey David,

I'm not sure what you mean by HiDPI displays - we have a Mac mini that's connected to an Acer monitor via HDMI to DVI converter.  I doubt that the display has a very high DPI to it, but I can try to figure out the exact value if you think it would help.  This also occurs when I use VNC with my laptop which I don't think has a very high DPI either (but then again I'm not sure what 'high' is here).

(In reply to David Baron [:dbaron] (needinfo? me) (UTC-8) from comment #15)
> Could this be specific to HiDPI displays?  (Are the people who see it on Mac
> all using such displays?)

Comment 17

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

No, don't bother -- effectively, it's retina displays.

I was just trying to think about why this would happen on Mac but not other platforms.

Comment 18

[Hani Yacoub, Desktop QA]

I couldn't manage to reproduce this issue. Since the bug was logged 9 years ago, most likely isn't reproducible anymore for the reporter also.
Closing this bug as resolved: Worksforme.