51968 - align=center should override text-indent

Status: RESOLVED INVALID

(Core :: Layout, defect, P3)

Description

[Hallvard B Furuseth]

From Bugzilla Helper:
User-Agent: Mozilla/4.75 [en] (X11; U; SunOS 5.7 sun4u)
BuildID:    2000090809

<p align="center" style="text-indent:5em">
does not center 1st line; it is pushed 5em
to the left.  That's not serious for a 1-line
paragraph, but looks bad with multiline
paragraphs with short text per line.

Reproducible: Always
Steps to Reproduce:
1. Visit the upcoming attachment.  It contains:

<style><!-- p { text-indent:5em; } --></style>
<p>Paragraphs<br>have text-indent=5em:</p>

<p align="center">P ALIGN=CENTER<br>1<br>2</p><br>
<center><p>P inside CENTER<br>1<br>2</p></center><br>
<p><center>CENTER inside P<br>1<br>2</center></p><br>


Actual Results:  1st line is not centered with
<p align=center>xxx or <center><p>xxx.
It _is_ centered with <p><center>xxx.

Expected Results:  If the user said <p align=center>, he almost
certainly meant the whole paragraph to be
centered.  I'm not quite as sure of <center><p>.

Comment 1

[Hallvard B Furuseth]

Attachment: HTML example of center+text-align

Comment 2

[Chris Petersen]

Problem may have been fixed. Please check report with latest nightly build from
mozilla.org

Comment 3

[Hallvard B Furuseth]

Sorry, not fixed in build ID 2000100209 (sun-solaris).

BTW, my "<CENTER> inside <P>" text is silly, I forgot
that <CENTER> terminates <P>.  It shows how I think the
other paragraphs _should_ look, though.

Comment 4

[buster]

P3 bugs will not make NS6 RTM.  Milestone->future.

Comment 5

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

What's the correct behavior here, according to CSS?  I suspect we may be right
already, but if we are, should the spec be changed?

Comment 6

[Hallvard B Furuseth]

That seems to be true for a lot of res/quirk.css as well.

Comment 7

[Asa Dotzler [:asa]]

so this bug should either be fixed (if we are broken), Resolved Invalid (not a
bug in mozilla) or Won't Fix (a bug in mozilla but we like it that way). Either
way, there is no reason for this bug to be Unconfirmed.  Setting bug status to New

Comment 8

[aldaniel]

*** Bug 64028 has been marked as a duplicate of this bug. ***

Comment 9

[Shirley Liu]

Problem can be reproduced on Netscape6.01A,Sparc Solaris2.8.
It works with Mozilla build 2001040904 on PC win2000.

Comment 10

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

I guess it's not clear to me what this bug is.  Are you saying that the
first line is 5em from the left edge, or that it's 2.5em to the right of
being centered?  I think the latter is correct (rather than perfectly
centered).

Please test on current Mozilla build on Solaris rather than Netscape 6.01,
although I doubt this is platform specific...

Comment 11

[Hallvard B Furuseth]

It's 2.5em to the right of being centered.
This is build ID 2001041722 on Sun-Solaris.

I agree 2.5em is better than 5em, however that still looks silly.
I've NEVER noticed a case of center where the author intended 1st line to
be indented as well.  The problem is that text-indent and alignment both
specify the same thing: how to format the paragraph.  In particular when
the author centers a few paragraphs and forgets that text-indent exists,
while the user's default style sheet has
	p { text-indent:something; margin-<top/bottom>: 0; }
I have that, and I've tried to add tricks like
	p[align="center"] { text-indent:inherit; }
but I can't get it to work properly.

So I think text-indent should not be applied to centered text, however
it might make sense to add it to both left and right margin of the first
line.   That way 1st line is not pushed to the left of normally
text-indented lines, but it still stays centered.

(What the CSS spec says I don't know.  If it says Mozilla is correct then
I suspect the problem didn't occur to the people who wrote it.)

I admit this scheme would do the Wrong Thing if several centered paragraphs
follow each other and need to be possible to tell apart, but (a) I've
never seen that, and (b) I'm not convinced that the reader would "see"
the "indented centered" line as a paragraph start anyway, unless he
was aware of this quirk and knew what to look for.

Comment 12

[Hixie (not reading bugmail)]

You want:
   p[align="center"] { text-indent: 0; }

Comment 13

[Hallvard B Furuseth]

No, "0" vs. "inherit" isn't the problem.

First, it isn't enough.
   p[align="center"] { text-indent: 0; }
doesn't override <p style="text-align:center">, and 
   p[text-align="center"] { ... }
doesn't either.  (Is that a bug?)
Nor does it catch text-indent in <center><p>...</p></center>, so I must add
   center p { text-indent: 0; }
as well.  And of course
   *[align="center"] p { text-indent: 0; }
to catch e.g. <table align=center>.  And maybe I'd better add
   p[align="left"] { text-indent: <whatever>; }
in case something explicitly left-justifies paragraph inside a centered
dingbat (e.g. in a table with centered elements)...

Second, I don't think _I_ should have to add it.  Unless you mean "I" as
in "user who can modify the source".  Like I said, I've never noticed a
case where the current behaviour is what the author or user desired.
I don't think anyone who puts p{text-indent:..} in his style sheet, should
have to add a bunch of hacks like the above.

Comment 14

[Kevin McCluskey (gone)]

Build reassigning Buster's bugs to Marc.

Comment 15

[Greg K.]

Verifying behavior using FizzillaCFM/2002070913. 'text-indent' is still applied
paragraphs that also have 'text-align: center'. Setting All/All.

Comment 16

[Boris Zbarsky [:bzbarsky]]

To Ian; we're currently rendering per the CSS spec, as far as one can....
Perhaps the spec should specify that when text-align:center text-indent is not
applied?  Or otherwise specify the rendering in a sane fashion?

In any case, please let me know what changes, if any, are needed in Mozilla's
behavior.

Comment 17

[Hixie (not reading bugmail)]

This was discussed in a CSSWG telecon recently. I doubt that the spec will be
changed away from what it says today. This bug is therefore invalid. Although I
agree that the suggestion makes sense, compliance and interoperability is more
important an the long run.