1474663 - Implement contain:size for nsColumnSetFrame (for multicol)

Status: RESOLVED FIXED

(Core :: Layout, enhancement, P2)

Description

[Daniel Holbert [:dholbert]]

We're importing a "contain:size" web platform test in bug 1474611 that fails with our current contain:size implementation.  The test is currently available here:
https://hg.mozilla.org/try/raw-file/2d4edac5f564/testing/web-platform/tests/css/css-contain/contain-size-multicol.html

(this will expire with the try build in a week, but at that point the test will probably live in mozilla-central)

I suspect it requires changes to nsColumnSetFrame.

Comment 1

[Morgan Reschenberg [:morgan]]

Attachment: Bug 1474663 - Implement contain:size for nsColumnSetFrame.

Review commit: https://reviewboard.mozilla.org/r/256652/diff/#index_header
See other reviews: https://reviewboard.mozilla.org/r/256652/

Comment 2

[Daniel Holbert [:dholbert]]

Comment on attachment 8991707 [details]
Bug 1474663 - Implement contain:size for nsColumnSetFrame.

https://reviewboard.mozilla.org/r/256652/#review263488

::: layout/generic/nsColumnSetFrame.cpp:497
(Diff revision 1)
> +  if (StyleDisplay()->IsContainSize()) {
> +    // if we are size contained and have no set width, we should behave
> +    // as if we have no children and have zero width
> +    iSize = 0;
> +    return iSize;
> +  }
>    if (mFrames.FirstChild()) {
>      iSize = mFrames.FirstChild()->GetMinISize(aRenderingContext);
>    }

Two things:

(1) I think we still need to take column-gap etc. into account here, right?

Really, I suspect you might want to change the "mFrames.FirstChild()" condition to have "&& !StyleDisplay()->IsContainSize()".  That's the specific spot where we take the child width into consideration.

(2) It's best not to mention "...and have no set width..." here.  These GetPref/MinISize functions aren't meant to care about set widths -- they're about our *intrinsic* widths that we get from contents & other properties.  So, probably better not to muddy things by implicitly suggesting that a fixed width might influence the behavior of this function.  Maybe reword as something like: "If we're size contained, then we disregard our contents when determining intrinsic size".

::: layout/generic/nsColumnSetFrame.cpp:548
(Diff revision 1)
> +  } else if (StyleDisplay()->IsContainSize()) {
> +    // if we are size contained and have no set width, we should behave
> +    // as if we have no children and have zero width
> +    colISize = 0;
>    } else if (mFrames.FirstChild()) {
>      colISize = mFrames.FirstChild()->GetPrefISize(aRenderingContext);
>    } else {
>      colISize = 0;
>    }

If you like, this could be simplified to "else if (mFrames.FirstChild() && !StyleDisplay()->IsContainSize())" -- that would make it closer to the logic that I'm envisioning that GetMinISize should end up with.

But it seems fine this way too (though at least the comment here needs a tweak, for the same reasons noted above in GetMinISize)

::: layout/generic/nsColumnSetFrame.cpp:922
(Diff revision 1)
>                                         aConfig.mComputedBSize);
>      } else {
>        contentSize.BSize(wm) = aConfig.mComputedBSize;
>      }
> +  } else if (aReflowInput.mStyleDisplay->IsContainSize()) {
> +    // If we are intrinsically sized, but are size contianed

s/contianed/contained,/

(fix typo and add comma)

::: layout/generic/nsColumnSetFrame.cpp:924
(Diff revision 1)
> +    // should be zero or minBSize if specified.
> +      contentSize.BSize(wm) =
> +        aReflowInput.ApplyMinMaxBSize(0);

Indentation is odd here:
 - "contentSize" should be deindented
 - I don't think you need to bother linebreaking after the "="... that expression looks like it fits on a single line.

Comment 3

[Daniel Holbert [:dholbert]]

A few more notes:
 - The patch seems to be removing a file called "myreftest.list", but that file doesn't exist in mozilla-central. It looks like it's actually created in the parent commit in your local tree, I think (the commit for bug 1470176) -- this one (found in mozreview): https://reviewboard-hg.mozilla.org/gecko/rev/c005c3852db6

 - If you want to test the GetMinISize behavior (probably a good idea), you can do so by giving the element width:auto and making it a child of a width:0 block.  Or, simpler, you can explicitly style it with "width: -moz-min-content" (though you probably can't use that in w3c-submitted tests -- "min-content" is the official spelling, though we don't support that spelling yet).

Comment 4

[Morgan Reschenberg [:morgan]]

(In reply to Daniel Holbert [:dholbert] from comment #3)
>  - If you want to test the GetMinISize behavior (probably a good idea), you
> can do so by giving the element width:auto and making it a child of a
> width:0 block.  Or, simpler, you can explicitly style it with "width:
> -moz-min-content" (though you probably can't use that in w3c-submitted tests
> -- "min-content" is the official spelling, though we don't support that
> spelling yet).

I tested this locally with the div-within-a-div strategy; do you think its a good idea to include that on the reftests or should I leave it off and stick with the boilerplate stuff I have so far? (in addition to testing column-gaps/column-widths)

Comment 5

[Morgan Reschenberg [:morgan]]

Comment on attachment 8991707 [details]
Bug 1474663 - Implement contain:size for nsColumnSetFrame.

Review request updated; see interdiff: https://reviewboard.mozilla.org/r/256652/diff/1-2/

Comment 6

[Daniel Holbert [:dholbert]]

It'd be a good idea to test it one way or another (for all frame types), whether that's part of the main test boilerplate, or a separate tiny boilerplateish "contain-size-$TYPE-intrinsic-001.html"  (for "$TYPE" == block, flex, button, multicol, etc), or something else.

Up to you whether you'd like to extend the boilerplate (for multicol) here vs. do that in a separate bug/patch in a more comprehensive way.

Comment 7

[Morgan Reschenberg [:morgan]]

(In reply to Daniel Holbert [:dholbert] from comment #6)
> It'd be a good idea to test it one way or another (for all frame types),
> whether that's part of the main test boilerplate, or a separate tiny
> boilerplateish "contain-size-$TYPE-intrinsic-001.html"  (for "$TYPE" ==
> block, flex, button, multicol, etc), or something else.
> 
> Up to you whether you'd like to extend the boilerplate (for multicol) here
> vs. do that in a separate bug/patch in a more comprehensive way.

Sounds good. I separated the reftests into two files -002, -003 to keep them under ~400px height. The -002 test has all the main boilerplate tests and -003 has multicol specific ones, so I added it in the latter. If I've got time later I'll spin off a bug to put the same ones into the other types too.

Comment 8

[Daniel Holbert [:dholbert]]

Comment on attachment 8991707 [details]
Bug 1474663 - Implement contain:size for nsColumnSetFrame.

https://reviewboard.mozilla.org/r/256652/#review263550

::: layout/reftests/w3c-css/submitted/contain/contain-size-multicol-003.html:21
(Diff revision 2)
> +  .col-width {
> +    column-width: 20px;
> +  }
> +  .col-gap {
> +    column-gap: 5px;
> +  }
> +  .flexBaselineCheck {
> +    display: flex;
> +    align-items: baseline;
> +  }
> +  </style>
> +</head>
> +<body>
> +  <!--CSS Test: A size-contained multicol element with specified column-width should render at given width and zero height regardless of content.-->
> +  <div class="contain col-width"><div class="innerContents">inner</div></div>
> +  <br>
> +
> +  <!--CSS Test: A size-contained multicol element with specified column-gap should render at given width and zero height regardless of content.-->
> +  <div class="contain col-gap"><div class="innerContents">inner</div></div>
> +  <br>

For the col-width and col-gap classes to have visible effects, we really need to test them on *intrinsically sized* multicol elements.

(Right now, this test is just applying them on multicol elements that are *already sized to fill their containing block* (by virtue of how width:auto works on block-level elements), so we effectively throw away the results of GetPrefISize / GetMinISize on these elements & their  column-gap and column-width values don't get a chance to make a visible difference in sizing. So we're not actually testing their effects, in practice.)

Really, I think you should add a set of classes:
 .min { width: -moz-min-content; }
 .max { width: -moz-max-content; }

...and you should double each of these first two divs, to test each one with maxContent and with minContent, e.g.:
  <div class="contain min col-width"><div class="innerContents">inner</div></div>
  <div class="contain max col-width"><div class="innerContents">inner</div></div>

(And for this to work in our w3c-submitted reftests, we also need to update the sync-tests.sh script so it'll automatically strip out the "-moz-" prefixes when we submit these upstream.)

::: layout/reftests/w3c-css/submitted/contain/contain-size-multicol-003.html:34
(Diff revision 2)
> +  <!--CSS Test: A size-contained multicol element with specified column-width should render at given width and zero height regardless of content.-->
> +  <div class="contain col-width"><div class="innerContents">inner</div></div>
> +  <br>
> +
> +  <!--CSS Test: A size-contained multicol element with specified column-gap should render at given width and zero height regardless of content.-->
> +  <div class="contain col-gap"><div class="innerContents">inner</div></div>
> +  <br>
> +
> +  <!--CSS Test: A size-contained multicol element with specified column-gap and column-width should render at total width and zero height regardless of content.-->

The phrases "render at given width" and "render at total width" are confusing here, because there is no given width (and "total width" isn't a common / well-defined term).

However, per my other note here, these chunks probably want to change anyway, so I imagine these comments will probably be replaced for that change already.

::: layout/reftests/w3c-css/submitted/contain/contain-size-multicol-003.html:52
(Diff revision 2)
> +  <div class="flexBaselineCheck">
> +  outside before<div class="contain"><div class="innerContents">inner</div></div>outside after
> +  </div>
> +  <br>
> +
> +  <!--CSS Test: A size-contained multicol element with auto width in a zero-width div should not display-->

"should not display" isn't the right term here (that sounds like display:none which would prevent any painting including border, for example).

I think you want to say "...should end up with a zero-sized content box"

::: layout/reftests/w3c-css/submitted/contain/contain-size-multicol-003.html:52
(Diff revision 2)
> +  <!--CSS Test: A size-contained multicol element with auto width in a zero-width div should not display-->
> +  <div style="width:0;">
> +    <div class="contain" style="width:auto;">
> +      <div class="innerContents">inner</div>
> +    </div>
> +  </div>

"should not display" isn't the right term here (that sounds like display:none which would prevent any painting including border, for example).

However, I think you probably want to just get rid of this whole width:0 subtree, in favor of the min-content/max-content stuff noted above -- because really, we need to be testing this tiny-sizing-behavior *in combination* with column-gap and column-width (and right now, this is testing the min-sizing behavior in the absence of those things).

::: layout/reftests/w3c-css/submitted/contain/reftest.list:25
(Diff revision 2)
>  == contain-size-multicol-001.html contain-size-multicol-001-ref.html
> -== contain-size-select-001-.html contain-size-select-001-ref.html
> +== contain-size-multicol-002.html contain-size-multicol-002-ref.html
> +== contain-size-multicol-003.html contain-size-multicol-003-ref.html

Looks like the patch expects to be stomping on a "contain-size-select-001" test in reftest.list right now, which I assume is added in one of your other patches.

That line doesn't exist in mozilla-central right now, so in order for this to autoland cleanly, you probably need to rebase this patch to be a direct child of the latest commit you've got from mozilla-central.

(I think that'll only cause bitrot in this reftest.list file, which you should be able to fix up manually by just copypasting these 3 lines in.)

Comment 9

[Daniel Holbert [:dholbert]]

Comment on attachment 8991707 [details]
Bug 1474663 - Implement contain:size for nsColumnSetFrame.

https://reviewboard.mozilla.org/r/256652/#review263550

> "should not display" isn't the right term here (that sounds like display:none which would prevent any painting including border, for example).
> 
> I think you want to say "...should end up with a zero-sized content box"

(oops, sorry, I ended up with two versions of this nit --> Dropping the more-stale version that I wrote first.)

Comment 10

[Daniel Holbert [:dholbert]]

BTW, I filed https://bugs.chromium.org/p/chromium/issues/detail?id=863454 on Chrome for incorrectly calculating the intrinsic sizes of contain:size multicol elements.

Comment 11

[Morgan Reschenberg [:morgan]]

Comment on attachment 8991707 [details]
Bug 1474663 - Implement contain:size for nsColumnSetFrame.

https://reviewboard.mozilla.org/r/256652/#review263550

> "should not display" isn't the right term here (that sounds like display:none which would prevent any painting including border, for example).
> 
> However, I think you probably want to just get rid of this whole width:0 subtree, in favor of the min-content/max-content stuff noted above -- because really, we need to be testing this tiny-sizing-behavior *in combination* with column-gap and column-width (and right now, this is testing the min-sizing behavior in the absence of those things).

After adding in the tests for min-/max-{count, width} and min-/max-{count, width, gap}, do you think it's worth testing min-/max{count} as well? This test was originally to make sure a column object that has no attributes that contribute to its height (i.e. one which is intrinsically sized) also renders correctly.

> Looks like the patch expects to be stomping on a "contain-size-select-001" test in reftest.list right now, which I assume is added in one of your other patches.
> 
> That line doesn't exist in mozilla-central right now, so in order for this to autoland cleanly, you probably need to rebase this patch to be a direct child of the latest commit you've got from mozilla-central.
> 
> (I think that'll only cause bitrot in this reftest.list file, which you should be able to fix up manually by just copypasting these 3 lines in.)

Rebased to today's moz-central copy, but I'll keep an eye on this and rebase before landing once it's r+'d. This file always seems to generate merge conflicts :(

Comment 12

[Morgan Reschenberg [:morgan]]

Comment on attachment 8991707 [details]
Bug 1474663 - Implement contain:size for nsColumnSetFrame.

https://reviewboard.mozilla.org/r/256652/#review263550

> The phrases "render at given width" and "render at total width" are confusing here, because there is no given width (and "total width" isn't a common / well-defined term).
> 
> However, per my other note here, these chunks probably want to change anyway, so I imagine these comments will probably be replaced for that change already.

I'm removing the tests that only have "contain col-gap" since per IRC discussion yesterday, they don't really make sense to test on their own with min-/max-.

Regarding descriptions on the other test cases: should I describe the min height as "... should render at column-width * 1 ..." and the max height as "... should render at (column-width * column-count) + gaps ..." or is there a (maybe less formulaic) description to give? If these are going in w3 I want to make sure they have decent notation since the behaviour of contain:size on columns is more complex than other objects. 

(That said, though, the wording above seems impercise to me because "gaps" (col-gap) isn't explicitly specified in all cases but has a default inherent value which still contributes to the width in the max-{count, width} case.)

Comment 13

[Daniel Holbert [:dholbert]]

Comment on attachment 8991707 [details]
Bug 1474663 - Implement contain:size for nsColumnSetFrame.

https://reviewboard.mozilla.org/r/256652/#review263550

> I'm removing the tests that only have "contain col-gap" since per IRC discussion yesterday, they don't really make sense to test on their own with min-/max-.
> 
> Regarding descriptions on the other test cases: should I describe the min height as "... should render at column-width * 1 ..." and the max height as "... should render at (column-width * column-count) + gaps ..." or is there a (maybe less formulaic) description to give? If these are going in w3 I want to make sure they have decent notation since the behaviour of contain:size on columns is more complex than other objects. 
> 
> (That said, though, the wording above seems impercise to me because "gaps" (col-gap) isn't explicitly specified in all cases but has a default inherent value which still contributes to the width in the max-{count, width} case.)

RE descriptions, I tend to think we should avoid being overly specific/verbose on the expected result -- perhaps it's easiest & cleanest to just say "...should size as if it had no contents"?

Really, that's all these tests are verifying, typically, since the reference case matches the testcase except with "contain:size" swapped for absence-of-content.  (We could say "should render as [formula]", but really, the tests aren't precisely validating that.)

> After adding in the tests for min-/max-{count, width} and min-/max-{count, width, gap}, do you think it's worth testing min-/max{count} as well? This test was originally to make sure a column object that has no attributes that contribute to its height (i.e. one which is intrinsically sized) also renders correctly.

Up to you, RE adding that scenario. I don't think there's any special rendering behavior in that "min/max-content combined with column-count" case, so it probably wouldn't get us much increased strictness...  But perhaps it might get us a bit better code-coverage; not sure.

Comment 14

[Morgan Reschenberg [:morgan]]

Comment on attachment 8991707 [details]
Bug 1474663 - Implement contain:size for nsColumnSetFrame.

Review request updated; see interdiff: https://reviewboard.mozilla.org/r/256652/diff/2-3/

Comment 15

[Daniel Holbert [:dholbert]]

Comment on attachment 8991707 [details]
Bug 1474663 - Implement contain:size for nsColumnSetFrame.

https://reviewboard.mozilla.org/r/256652/#review263804

::: layout/generic/nsColumnSetFrame.cpp:499
(Diff revision 3)
> +    // We want to ignore this in the case that we're size contained
> +    // becuase our children should not contribute to our

typo: s/becuase/because/

(please fix both copies of this comment -- in both GetMinISize and GetPrefISize)

::: layout/reftests/w3c-css/submitted/contain/contain-size-multicol-002-ref.html:10
(Diff revision 3)
> +  <title>CSS Reftest Reference</title>
> +  <link rel="author" title="Morgan Rae Reschenberg" href="mailto:[e-mail]">
> +  <style>
> +  .basic {
> +    border: 1em solid green;
> +    columns: 3;

Throughout all 4 HTML files here, let's specify this as "column-count" rather than "columns".

(Reason: the "columns" property is actually a shorthand that specifies/resets **both** the column-count and column-width properties at once. So right now, for things like <div class="basic width-ref">, there's a fragile dependency on the precise ordering of these CSS rules, because each CSS rule specifies a different value for the column-width property, and whichever one comes last "wins". If you fix this to use "column-count", then that resolves this problem because this rule will stop inadvertantly resetting column-width, and the other one will unambiguously be the only thing to set column-width, and there'll be no more conflict/fragile-ordering-dependency.)

::: layout/reftests/w3c-css/submitted/contain/contain-size-multicol-003-ref.html:10
(Diff revision 3)
> +  <title>CSS Reftest Reference</title>
> +  <link rel="author" title="Morgan Rae Reschenberg" href="mailto:[e-mail]">
> +  <style>
> +  .basic {
> +    border: 1em solid green;
> +    columns:3;

Nit: Add a space before "3" here, so that this doesn't show up as difference between the tescase & reference case here when comparing them in a diff tool.

(And as noted for another file above, let's use "column-count" rather than "columns" here.)

::: layout/reftests/w3c-css/submitted/contain/contain-size-multicol-003-ref.html:12
(Diff revision 3)
> +  .col-width-ref {
> +    column-width: 20px;
> +  }
> +  .col-gap-ref {
> +    column-gap: 5px;

Minor nit -- no changes needed here or in existing tests, just mentioning for future reference where it's handy:

I'd suggest leaving off these "-ref" classname suffixes in general, except in scenarios where the classname would be misleading/confusing. (e.g. If the testcase's classname were .someFeature, and "someFeature" is a feature that you're *not using* in the reference case because you're creating a mocked-up version of it, then .someFeature-ref would be good & clearer name in the reference case.)

With these suffixes left off, then testcase/reference file pairs will end up being easier to compare in a merge tool, to narrow in on specifically what's being tested (which will show up as the only textual diff between the files).  And it makes your life a bit easier, too, since it's one fewer thing that you have to rename throughout one of the files when writing it.

(Again, this is very minor & please feel 100% free to leave this as-is for these tests; just mentioning as a suggestion for later tests.)

Comment 16

[Daniel Holbert [:dholbert]]

(This is r+ with feedback addressed, but do remember to hold off on requesting landing here until bug 1475645 is taken care of.)

Comment 17

[Morgan Reschenberg [:morgan]]

Comment on attachment 8991707 [details]
Bug 1474663 - Implement contain:size for nsColumnSetFrame.

Review request updated; see interdiff: https://reviewboard.mozilla.org/r/256652/diff/3-4/

Comment 18

[Morgan Reschenberg [:morgan]]

(oops, meant to mark bug 1475645 as resolved and this with checkin keyword; got the two tabs mixed up)

PR on 1475645 went through and got merged, so these changes are okay to land now :) Did a rebase to this morning's moz-central before pushing changes on feedback above.

Comment 19

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/4aa2f74bad5b
Implement contain:size for nsColumnSetFrame. r=dholbert

Comment 20

[Bogdan Tara[:bogdan_tara | bogdant]]

Backed out changeset 4aa2f74bad5b (bug 1474663) for contain size failures.

Backout link: https://hg.mozilla.org/integration/autoland/rev/0287f9144ec2f38ae2f8d9dd8f0dfc5745133514

Push with failures: https://treeherder.mozilla.org/#/jobs?repo=autoland&revision=4aa2f74bad5b77dd8f8c7e6e0e9eacafd64ef95d

Failure log: https://treeherder.mozilla.org/logviewer.html#?job_id=188410894&repo=autoland

[task 2018-07-16T17:48:42.476Z] 17:48:42     INFO - REFTEST TEST-START | file:///builds/worker/workspace/build/tests/reftest/tests/layout/reftests/w3c-css/submitted/contain/contain-size-multicol-003.html == file:///builds/worker/workspace/build/tests/reftest/tests/layout/reftests/w3c-css/submitted/contain/contain-size-multicol-003-ref.html
[task 2018-07-16T17:48:42.484Z] 17:48:42     INFO - REFTEST INFO | SET PREFERENCE pref(layout.css.contain.enabled,true)
[task 2018-07-16T17:48:42.485Z] 17:48:42     INFO - REFTEST TEST-LOAD | file:///builds/worker/workspace/build/tests/reftest/tests/layout/reftests/w3c-css/submitted/contain/contain-size-multicol-003.html | 0 / 1 (0%)
[task 2018-07-16T17:48:42.507Z] 17:48:42     INFO - ++DOMWINDOW == 4 (0x7f69de427800) [pid = 1209] [serial = 4] [outer = 0x7f69f7798600]
[task 2018-07-16T17:48:42.639Z] 17:48:42     INFO - ### XPCOM_MEM_BLOAT_LOG defined -- logging bloat/leaks to /tmp/tmpJ675yv.mozrunner/runreftest_leaks_tab_pid1235.log
[task 2018-07-16T17:48:42.762Z] 17:48:42     INFO - REFTEST INFO | drawWindow flags = DRAWWINDOW_DRAW_CARET | DRAWWINDOW_DRAW_VIEW | DRAWWINDOW_USE_WIDGET_LAYERS; window size = 800,1000; test browser size = 800,1000
[task 2018-07-16T17:48:42.940Z] 17:48:42     INFO - REFTEST INFO | RESTORE PREFERENCE pref(layout.css.contain.enabled,false)
[task 2018-07-16T17:48:42.942Z] 17:48:42     INFO - REFTEST INFO | SET PREFERENCE pref(layout.css.contain.enabled,true)
[task 2018-07-16T17:48:42.943Z] 17:48:42     INFO - REFTEST TEST-LOAD | file:///builds/worker/workspace/build/tests/reftest/tests/layout/reftests/w3c-css/submitted/contain/contain-size-multicol-003-ref.html | 0 / 1 (0%)
[task 2018-07-16T17:48:42.985Z] 17:48:42     INFO - ++DOMWINDOW == 5 (0x7f69de424800) [pid = 1209] [serial = 5] [outer = 0x7f69f7798600]
[task 2018-07-16T17:48:43.029Z] 17:48:43     INFO - Couldn't convert chrome URL: chrome://branding/locale/brand.properties
[task 2018-07-16T17:48:43.421Z] 17:48:43     INFO - REFTEST TEST-UNEXPECTED-FAIL | file:///builds/worker/workspace/build/tests/reftest/tests/layout/reftests/w3c-css/submitted/contain/contain-size-multicol-003.html == file:///builds/worker/workspace/build/tests/reftest/tests/layout/reftests/w3c-css/submitted/contain/contain-size-multicol-003-ref.html | image comparison, max difference: 255, number of differing pixels: 7319
[task 2018-07-16T17:48:43.421Z] 17:48:43     INFO - REFTEST   IMAGE 1 (TEST): data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAyAAAAPoC
[task 2018-07-16T17:48:43.421Z] 17:48:43     INFO - REFTEST   IMAGE 2 (REFERENCE): data:image/png;base64,iVBORw0KGgoAAA
[task 2018-07-16T17:48:43.422Z] 17:48:43     INFO - REFTEST INFO | Saved log: START file:///builds/worker/workspace/build/tests/reftest/tests/layout/reftests/w3c-css/submitted/contain/contain-size-multicol-003.html
[task 2018-07-16T17:48:43.424Z] 17:48:43     INFO - REFTEST INFO | Saved log: [CONTENT] Using browser remote=true
[task 2018-07-16T17:48:43.424Z] 17:48:43     INFO - 
[task 2018-07-16T17:48:43.426Z] 17:48:43     INFO - REFTEST INFO | Saved log: [CONTENT] OnDocumentLoad triggering AfterOnLoadScripts
[task 2018-07-16T17:48:43.427Z] 17:48:43     INFO - REFTEST INFO | Saved log: Initializing canvas snapshot
[task 2018-07-16T17:48:43.428Z] 17:48:43     INFO - REFTEST INFO | Saved log: DoDrawWindow 0,0,800,1000
[task 2018-07-16T17:48:43.429Z] 17:48:43     INFO - REFTEST INFO | Saved log: [CONTENT] RecordResult fired
[task 2018-07-16T17:48:43.430Z] 17:48:43     INFO - REFTEST INFO | Saved log: RecordResult fired
[task 2018-07-16T17:48:43.431Z] 17:48:43     INFO - REFTEST INFO | Saved log: START file:///builds/worker/workspace/build/tests/reftest/tests/layout/reftests/w3c-css/submitted/contain/contain-size-multicol-003-ref.html
[task 2018-07-16T17:48:43.432Z] 17:48:43     INFO - REFTEST INFO | Saved log: [CONTENT] OnDocumentLoad triggering AfterOnLoadScripts
[task 2018-07-16T17:48:43.433Z] 17:48:43     INFO - REFTEST INFO | Saved log: Initializing canvas snapshot
[task 2018-07-16T17:48:43.435Z] 17:48:43     INFO - REFTEST INFO | Saved log: DoDrawWindow 0,0,800,1000
[task 2018-07-16T17:48:43.436Z] 17:48:43     INFO - REFTEST INFO | Saved log: [CONTENT] RecordResult fired
[task 2018-07-16T17:48:43.437Z] 17:48:43     INFO - REFTEST INFO | Saved log: RecordResult fired
[task 2018-07-16T17:48:43.438Z] 17:48:43     INFO - REFTEST TEST-END | file:///builds/worker/workspace/build/tests/reftest/tests/layout/reftests/w3c-css/submitted/contain/contain-size-multicol-003.html == file:///builds/worker/workspace/build/tests/reftest/tests/layout/reftests/w3c-css/submitted/contain/contain-size-multicol-003-ref.html
[task 2018-07-16T17:48:43.439Z] 17:48:43     INFO - ++DOMWINDOW == 6 (0x7f69de562400) [pid = 1209] [serial = 6] [outer = 0x7f69f7798600]
[task 2018-07-16T17:48:43.441Z] 17:48:43     INFO - REFTEST INFO | RESTORE PREFERENCE pref(layout.css.contain.enabled,false)
[task 2018-07-16T17:48:43.442Z] 17:48:43     INFO - REFTEST INFO | Slowest test took 440ms (file:///builds/worker/workspace/build/tests/reftest/tests/layout/reftests/w3c-css/submitted/contain/contain-size-multicol-003.html)
[task 2018-07-16T17:48:43.443Z] 17:48:43     INFO - REFTEST INFO | Total canvas count = 0

Comment 21

[Morgan Reschenberg [:morgan]]

Comment on attachment 8991707 [details]
Bug 1474663 - Implement contain:size for nsColumnSetFrame.

https://reviewboard.mozilla.org/r/256652/#review264144

::: layout/reftests/w3c-css/submitted/contain/contain-size-multicol-003-ref.html:10
(Diff revision 4)
> +  <title>CSS Reftest Reference</title>
> +  <link rel="author" title="Morgan Rae Reschenberg" href="mailto:[e-mail]">
> +  <style>
> +  .basic {
> +    border: 1em solid green;
> +    columns-count: 3;

s/columns-count/column-count

Comment 22

[Morgan Reschenberg [:morgan]]

Comment on attachment 8991707 [details]
Bug 1474663 - Implement contain:size for nsColumnSetFrame.

Review request updated; see interdiff: https://reviewboard.mozilla.org/r/256652/diff/4-5/

Comment 23

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/e9cd9d73e5a8
Implement contain:size for nsColumnSetFrame. r=dholbert

Comment 24

[Tiberius Oros[:tiberius_oros]]

https://hg.mozilla.org/mozilla-central/rev/e9cd9d73e5a8

Comment 25

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

wpt test import at https://github.com/web-platform-tests/wpt/pull/12042

Comment 26

[Cristian Comorasu, QA [:ccomorasu], Releae Desktop QA - Please contact [e-mail]]

Marking this as qe-verify -, per comment 25.