12304 - incorrect overconstraint handling for rtl, absolute

Status: VERIFIED INVALID

(Core :: Layout, defect, P3)

Description

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

This is the second of two bugs for my finally half-completed test for CSS2
10.3.7 (widths and margins of absolutely positioned non-replaced elements).  The
first is bug 12302.

DESCRIPTION:  There are two cases in the above URL that are severly incorrect.
These are a1b1c1d0e1f1 and a1b1c1d1e1f1.  This means you are incorrectly
handling the overconstraint as described in the last rule (rule 7) of section
10.3.7.  You should be ignoring margin-left when direction is rtl, rather than
ignoring margin-right as you now are.

STEPS TO REPRODUCE:  Look at the above URL.

ACTUAL RESULTS:  The two aforementioned cases touch the left blue edge.

EXPECTED RESULTS:  They should, respectively, touch the right red edge (but see
bug 12302) and right blue edge.

DOES NOT WORK CORRECTLY ON:
 * Linux, apprunner, 1999-08-20-13-M10

Comment 1

[troy]

I don't see why you think a1b1c1d0e1f1 is overconstrained. The only thing I see
that's 'auto' is 'right'?

In the case of a1b1c1d1e1f1 I don't see any properties set to 'auto'?

Comment 2

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

"Overconstrained" is when nothing is auto after step 4.  In the first case,
right becomes non-auto in step 2 because direction is rtl.  In the second case,
nothing is auto to begin with.  These cases trigger step 7, which should set
margin-right to auto rather than margin-left.

Comment 3

[troy]

I had code in there at one point to deal with the case where nothing was 'auto',
but I ended up removing it because that caused me to fail another test you had.
Unfortunately I don't remember the details. I guess I'll add it back and see
what breaks...

Comment 4

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

ALSO DOES NOT WORK CORRECTLY ON:
 * Windows, apprunner, 1999-08-24-09-M10

Marking All/All

Comment 5

[Hixie (not reading bugmail)]

Migrating from {css2} to css2 keyword. The {css1}, {css2}, {css3} and {css-moz}
radars should now be considered deprecated in favour of keywords.
I am *really* sorry about the spam...

Comment 6

[Christine Hoffman]

Adding 'beta1' keyword. Although to date, it is not clear whether the 'right' 
CSS2 property is supported in beta1 (clarification has been requested), the 
property falls within the general CSS2 'absolute positioning' area, which should 
be supported by beta1. 

Comment 7

[troy]

It's CSS2 and so it's clear it should not be something required for beta. Plus, 
the W3C Style committee is still resolving how this should all work

Please don't change my bugs. Only David Baron and Ian are allowed to muck with 
my bugs

Comment 8

[Christine Hoffman]

'Beta1' discussion taken offline. If there are specific requirements as to who 
can input to this bug, please state clearly in the bug. Thanks.

Comment 9

[troy]

Marking this INVALID, becaus it isn't applicable now that I have implemented the 
new absolute positioning rules.

Comment 10

[Christine Hoffman]

I'm not sure why this is invalid unless there are new rules different from the 
CSS2 spec. I am attaching a new testcase where the values are over-constrained 
in an absolutely positioned div and the direction is 'rtl'. In this case, per 
the spec, the 'left' value is to be ignored and then the left value is solved 
again. In this example, the 'left' value should resolve to '0'. What is left is 
a 50px left margin value, a 350px width value, a 25px right margin value and a 
25px right value.

Comment 11

[Christine Hoffman]

Attachment: reopening bug with new testcase to demonstrate problem

Comment 12

[Christine Hoffman]

Above testcase tested on Windows with build #04-26-11-m16

Comment 13

[troy]

There are new absolute positioning rules and they will be in the upcoming CSS2 
errata

Marking the bug as INVALID again. Please don't reopen it again. If you have a 
different test case, then that's a different problem and you should open up a 
new bug...

Comment 14

[Christine Hoffman]

This bug was reopened based on existing CSS2 specifications. I am unable to 
verify this bug as 'invalid' until I can verify the changes.

Comment 15

[Christine Hoffman]

Will marked this verified invalid, pending official changes to spec

Comment 16

[(no longer active)]

Mass-assigning the new rtl keyword to RTL-related (see bug 349193).

Comment 17

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/be0b05f6b43e
[components] Closes https://github.com/mozilla-mobile/android-components/issues/12304: Handle exceptions in SearchEngineReader.loadStream