85341 - The dir attribute of the p element is not working correctly.

Status: RESOLVED INVALID

(Core :: Layout: Text and Fonts, defect)

Description

[bsharma]

Build: 2001-06-07-13-0.9.1 (RedHat 7.0, Kernel 2.2)

See above testcase, and you will see that a dot (.) appears in the begining of 
the sentence where as it should be in the end for right to left alignment of the 
line.

Comment 1

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

According to HTML 4.0 section 8.2 (see esp. 8.2.2),
<URL: http://www.w3.org/TR/html4/struct/dirlang.html#h-8.2 >, the dir attribute
specifies the base direction for the block.

According to section 3.12 within chapter 3
(<URL: http://www.unicode.org/unicode/uni2book/ch03.pdf >) of the Unicode
standard (<URL: http://www.unicode.org/unicode/uni2book/u2.html >), P2 and P3 do
not apply since the embedding level of the paragraph is specified by a higher
level protocol (HTML), and this embedding level says that the default direction
of text in the paragraph is RTL and its embedding level is 1 (see BD3 and BD4).
   The embedding level of the strong LTR text within the paragraph is going to
be 2.  The period (.) has a weak character type (CS) according to Table 3-8. 
Since there are no explicit embedding levels within the paragraph, the entire
paragraph is a single level run, with sor R and eor R (X10).  In the weak type
resolution, according to W6, the CS changes to ON (other neutral).  In the
neutral type resolution, according to N2, the period gets embedding level 1. 
Then, according to I2, all the text in the sentence other than the period (which
is still considered a neutral character since N1 did not apply) is promoted to
embedding level 2 (ltr).  Then, according to L2, the ltr text is reversed twice
and the whole line is reversed once, leaving the period at the beginning with
the rest of the text in English order, following it.

Marking INVALID, although someone who knows the Unicode Bidi algorithm better
might want to check that reasoning.